[{"id": 605128, "type": "GenericWebPage", "mime_type": "text/html", "url": "https://aistandardshub.org/ai-standards/standard-for-xai-explainable-artificial-intelligence-for-achieving-clarity-and-interoperability-of-ai-systems-design/", "slug": "aistandardshub", "title": "Standard for XAI \u2013 eXplainable Artificial Intelligence - for Achieving Clarity and Interoperability of AI Systems Design - AI Standards Hub", "snippet": "Last updated: 18 Jul 2024 \u00b7 This standard defines mandatory and optional requirements and constraints that need to be satisfied for an AI method, algorithm, application or system to be recognized as explainable. Both partially explainable and fully or strongly explainable methods, algorithms ...", "image": null, "date_published": null, "media": [], "products": [], "source": {"id": 605128, "website_id": 46455, "url": "https://aistandardshub.org/ai-standards/standard-for-xai-explainable-artificial-intelligence-for-achieving-clarity-and-interoperability-of-ai-systems-design/", "title": "Standard for XAI \u2013 eXplainable Artificial Intelligence - for Achieving Clarity and Interoperability of AI Systems Design - AI Standards Hub", "snippet": "Last updated: 18 Jul 2024 \u00b7 This standard defines mandatory and optional requirements and constraints that need to be satisfied for an AI method, algorithm, application or system to be recognized as explainable. Both partially explainable and fully or strongly explainable methods, algorithms ...", "cleaned_text": null, "cleaned_text_custom_parser": null, "cleaned_text_custom_parser_2": null, "best_field": null, "enhanced": false, "date_published": null, "image": null, "num_tokens_oai_tokenizer": null, "num_tokens_gemini_tokenizer": null}, "favicon": "https://aistandardshub.org/wp-content/uploads/2022/10/cropped-AI-Standards-Hub_Web-icon_512x512px-e1666003931812-32x32.png", "state": "candidate", "enhanced": false, "answer": null, "valid_answer": false, "source_citation_index": 1, "user_web_page_id": null, "status": "not_processed", "reader": false, "rank": 1, "reason": "The source provides specific standards and requirements for explainable AI, which aligns directly with emerging standards for AI explainability."}, {"id": 605126, "type": "GenericWebPage", "mime_type": "text/html", "url": "https://www.academia.edu/103925580/Explainability_in_AI_Policies_A_Critical_Review_of_Communications_Reports_Regulations_and_Standards_in_the_EU_US_and_UK", "slug": "academia", "title": "(PDF) Explainability in AI Policies: A Critical Review of Communications, Reports, Regulations, and Standards in the EU, US, and UK", "snippet": "Public attention towards explainability of artificial intelligence (AI) systems has been rising in recent years to offer methodologies for human oversight. This has translated into the proliferation of research outputs, such as from Explainable AI,", "image": null, "date_published": null, "media": [], "products": [], "source": {"id": 605126, "website_id": 2714, "url": "https://www.academia.edu/103925580/Explainability_in_AI_Policies_A_Critical_Review_of_Communications_Reports_Regulations_and_Standards_in_the_EU_US_and_UK", "title": "(PDF) Explainability in AI Policies: A Critical Review of Communications, Reports, Regulations, and Standards in the EU, US, and UK", "snippet": "Public attention towards explainability of artificial intelligence (AI) systems has been rising in recent years to offer methodologies for human oversight. This has translated into the proliferation of research outputs, such as from Explainable AI,", "cleaned_text": null, "cleaned_text_custom_parser": null, "cleaned_text_custom_parser_2": null, "best_field": null, "enhanced": false, "date_published": null, "image": null, "num_tokens_oai_tokenizer": null, "num_tokens_gemini_tokenizer": null}, "favicon": null, "state": "candidate", "enhanced": false, "answer": null, "valid_answer": false, "source_citation_index": 2, "user_web_page_id": null, "status": "not_processed", "reader": false, "rank": 1, "reason": "This source reviews various communications, regulations, and standards on AI explainability across different regions, making it highly relevant to the query."}, {"id": 500041, "type": "GenericWebPage", "mime_type": "text/html", "url": "https://www.unesco.org/en/articles/how-iso-and-iec-are-developing-international-standards-responsible-adoption-ai", "slug": "unesco", "title": "How the ISO and IEC are developing international standards for the - UNESCO", "snippet": "The joint ISO (International Organization for Standardization) and IEC (International Electrotechnical Commission) committee on Artificial Intelligence (AI), SC 42, is playing a crucial role in developing international standards to facilitate the responsible adoption of AI technologies. SC 42 has an ...", "image": null, "date_published": null, "media": [], "products": [], "source": {"id": 500041, "website_id": 13584, "url": "https://www.unesco.org/en/articles/how-iso-and-iec-are-developing-international-standards-responsible-adoption-ai", "title": "How the ISO and IEC are developing international standards for the - UNESCO", "snippet": "The joint ISO (International Organization for Standardization) and IEC (International Electrotechnical Commission) committee on Artificial Intelligence (AI), SC 42, is playing a crucial role in developing international standards to facilitate the responsible adoption of AI technologies. SC 42 has an ...", "cleaned_text": null, "cleaned_text_custom_parser": null, "cleaned_text_custom_parser_2": null, "best_field": null, "enhanced": false, "date_published": null, "image": null, "num_tokens_oai_tokenizer": null, "num_tokens_gemini_tokenizer": null}, "favicon": "https://www.unesco.org/themes/custom/bunesco8/icon.svg", "state": "candidate", "enhanced": false, "answer": null, "valid_answer": false, "source_citation_index": 3, "user_web_page_id": null, "status": "not_processed", "reader": false, "rank": 1, "reason": "The source discusses the development of international standards for AI by ISO and IEC, crucial for understanding emerging standards for explainability."}, {"id": 605129, "type": "GenericWebPage", "mime_type": "text/html", "url": "https://aisel.aisnet.org/wi2023/77/", "slug": "aisnet", "title": "Artificial Intelligence Explainability Requirements of the AI Act and Metrics for Measuring Compliance", "snippet": "Explainability in artificial intelligence (AI) is crucial for ensuring transparency, accountability, and risk mitigation, thereby addressing digital responsibility, social, ethical and ecological aspects of information system usage. AI will be regulated in the European Union (EU) through the AI Act.", "image": null, "date_published": null, "media": [], "products": [], "source": {"id": 605129, "website_id": 39533, "url": "https://aisel.aisnet.org/wi2023/77/", "title": "Artificial Intelligence Explainability Requirements of the AI Act and Metrics for Measuring Compliance", "snippet": "Explainability in artificial intelligence (AI) is crucial for ensuring transparency, accountability, and risk mitigation, thereby addressing digital responsibility, social, ethical and ecological aspects of information system usage. AI will be regulated in the European Union (EU) through the AI Act.", "cleaned_text": null, "cleaned_text_custom_parser": null, "cleaned_text_custom_parser_2": null, "best_field": null, "enhanced": false, "date_published": null, "image": null, "num_tokens_oai_tokenizer": null, "num_tokens_gemini_tokenizer": null}, "favicon": "https://aisel.aisnet.org/favicon.ico", "state": "candidate", "enhanced": false, "answer": null, "valid_answer": false, "source_citation_index": 4, "user_web_page_id": null, "status": "not_processed", "reader": false, "rank": 1, "reason": "The article discusses the educational aspects and compliance of AI explainability in the context of the AI Act, which directly pertains to evolving regulations."}, {"id": 500016, "type": "GenericWebPage", "mime_type": "text/html", "url": "https://standards.ieee.org/news/ieee-standards-commitment-to-advancing-ai-governance-includes-impactful-contributions-to-new-international-ai-standards-exchange/", "slug": "ieee", "title": "IEEE Standards Commitment to Advancing AI Governance Includes Impactful ...", "snippet": "IEEE strengthens global AI governance by contributing 100+ standards to the new International AI Standards Exchange, fostering innovation and interoperability", "image": null, "date_published": null, "media": [], "products": [], "source": {"id": 500016, "website_id": 33695, "url": "https://standards.ieee.org/news/ieee-standards-commitment-to-advancing-ai-governance-includes-impactful-contributions-to-new-international-ai-standards-exchange/", "title": "IEEE Standards Commitment to Advancing AI Governance Includes Impactful ...", "snippet": "IEEE strengthens global AI governance by contributing 100+ standards to the new International AI Standards Exchange, fostering innovation and interoperability", "cleaned_text": null, "cleaned_text_custom_parser": null, "cleaned_text_custom_parser_2": null, "best_field": null, "enhanced": false, "date_published": null, "image": null, "num_tokens_oai_tokenizer": null, "num_tokens_gemini_tokenizer": null}, "favicon": "https://standards.ieee.org/wp-content/themes/ieee-sa-theme/ieee_apple_touch_icon.png", "state": "candidate", "enhanced": false, "answer": null, "valid_answer": false, "source_citation_index": 5, "user_web_page_id": null, "status": "not_processed", "reader": false, "rank": 1, "reason": "The source explains the collaborative efforts of IEEE in advancing AI standards, which are essential for AI explainability and governance."}, {"id": 605127, "type": "GenericWebPage", "mime_type": "text/html", "url": "https://link.springer.com/article/10.1007/s12525-023-00644-5", "slug": "springer", "title": "Explainable artificial intelligence in information systems: A review of the status quo and future research directions | Electronic Markets", "snippet": "The quest to open black box artificial intelligence (AI) systems evolved into an emerging phenomenon of global interest for academia, business, and society and brought about the rise of the research field of explainable artificial intelligence (XAI). With its pluralistic view, information systems ...", "image": null, "date_published": null, "media": [], "products": [], "source": {"id": 605127, "website_id": 219, "url": "https://link.springer.com/article/10.1007/s12525-023-00644-5", "title": "Explainable artificial intelligence in information systems: A review of the status quo and future research directions | Electronic Markets", "snippet": "The quest to open black box artificial intelligence (AI) systems evolved into an emerging phenomenon of global interest for academia, business, and society and brought about the rise of the research field of explainable artificial intelligence (XAI). With its pluralistic view, information systems ...", "cleaned_text": null, "cleaned_text_custom_parser": null, "cleaned_text_custom_parser_2": null, "best_field": null, "enhanced": false, "date_published": null, "image": null, "num_tokens_oai_tokenizer": null, "num_tokens_gemini_tokenizer": null}, "favicon": "https://link.springer.com/oscar-static/img/favicons/darwin/apple-touch-icon-92e819bf8a.png", "state": "candidate", "enhanced": false, "answer": null, "valid_answer": false, "source_citation_index": 6, "user_web_page_id": null, "status": "not_processed", "reader": false, "rank": 2, "reason": "This source reviews the status of explainable AI in information systems, relevant to emerging standards and the state of the field."}, {"id": 333095, "type": "GenericWebPage", "mime_type": "text/html", "url": "https://link.springer.com/article/10.1007/s11063-025-11732-2", "slug": "springer", "title": "Recent Emerging Techniques in Explainable Artificial Intelligence to Enhance the Interpretable and Understanding of AI Models for Human - Neural Processing Letters", "snippet": "Recent advancements in Explainable Artificial Intelligence (XAI) aim to bridge the gap between complex artificial intelligence (AI) models and human understanding, fostering trust and usability in AI systems. However, challenges persist in comprehensively interpreting these models, hindering their widespread adoption. This study addresses these challenges by exploring recently emerging ...", "image": null, "date_published": null, "media": [], "products": [], "source": {"id": 333095, "website_id": 219, "url": "https://link.springer.com/article/10.1007/s11063-025-11732-2", "title": "Recent Emerging Techniques in Explainable Artificial Intelligence to Enhance the Interpretable and Understanding of AI Models for Human - Neural Processing Letters", "snippet": "Recent advancements in Explainable Artificial Intelligence (XAI) aim to bridge the gap between complex artificial intelligence (AI) models and human understanding, fostering trust and usability in AI systems. However, challenges persist in comprehensively interpreting these models, hindering their widespread adoption. This study addresses these challenges by exploring recently emerging ...", "cleaned_text": "AI explainability has many local use cases across different industries and applications [15]. The local use cases include healthcare, finance, retail and e-commerce, agriculture, education, transport and logistics, and legal and regulatory. These are succinctly presented in the Fig. 2 and discussed as follows:\n\n(i) Healthcare: Explainable Artificial Intelligence (XAI) plays a crucial role in disease diagnosis, AI models used in medical diagnosis can explain their decisions, helping doctors understand the reasoning behind a diagnosis. Ensuring explainability in AI for disease diagnosis is crucial for fostering trust, transparency, and acceptance of AI-based diagnostic systems within the healthcare sector. As highlighted in study by Asan et al. [11], when patients and healthcare providers comprehend the reasoning behind AI-generated diagnoses, they can make more informed decisions and place greater trust in the recommendations. Incorporating explainability into AI disease diagnosis not only assists healthcare professionals but also addresses concerns regarding safety, accountability, and ethical utilization of AI in healthcare [21]. It is imperative to strike a balance between model accuracy and interpretability to establish trust in the AI system and enable its effective integration into medical practice. According to Zhang et al. [170], eXplainable AI in healthcare focuses on several critical aspects, including ensuring transparent data and features, thereby enabling healthcare providers to comprehend the data inputs and variables influencing the diagnostic process. Secondly, XAI provides insights into feature importance and contribution, enabling clinicians to identify critical factors contributing to a diagnosis. Additionally, it offers visualization tools and graphs to represent complex data relationships visually, aiding in comprehension and decision-making. Moreover, XAI enhances model interpretability, allowing healthcare providers to understand how a diagnostic model arrives at its conclusions. Furthermore, Schoonderwoerd et al. [140] explained that XAI facilitates natural language explanations, which help communicate diagnostic results and treatment recommendations in a comprehensible manner to both clinicians and patients. XAI also supports treatment recommendations, comparative analysis between different diagnostic approaches, and patient education, thereby empowering healthcare professionals and patients with actionable insights and knowledge for informed decision-making in disease diagnosis and management. For instance, a recent study by Zhu [171] utilized computational intelligence with AI to cclassify brain images from MRI or CT scans, demonstrating the impact of dosage ratios on health over treatment duration, and diagnosing memory impairment in psychoactive substance users. The shift towards digital pathology allows pathologists to utilize AI-based computer programs for analyzing whole slide images, but many top-performing algorithms operate as 'black boxes,' leaving even developers uncertain about their decision-making processes [120]. This review emphasizes the necessity of explainability in digital pathology to prevent mistakes and ensure patient safety, introducing various explainable AI techniques that enhance the transparency of these algorithms. Furthermore, it advocates for the development of novel user interfaces that promote contextual understanding and interactive exploration, enabling medical experts to engage with AI results more effectively and maintain a critical human presence in the decision-making process.\n\n(ii) Finance: Weber et al. [163, 164], the explainability of AI in finance, particularly in credit scoring and risk assessment, is crucial for several reasons, including regulatory compliance, transparency, and trust-building with both lenders and borrowers. In credit scoring, AI models can explain why an applicant\u2019s credit score is calculated in a certain way, offering transparency to applicants and regulators while in risk assessment, explainable AI models can provide insights into the factors influencing risk assessment in financial investments. Explainability in AI for credit scoring and risk assessment helps lenders make responsible lending decisions and empowers borrowers to understand and improve their financial profiles [62]. Moreover, it assists regulatory bodies in ensuring that AI-driven credit decisions comply with legal and ethical standards, promoting fairness and accountability in the financial industry. Matcov [103] explained that XAI is instrumental in credit scoring and risk assessment within the finance sector, operating across several key dimensions like identifying feature importance, and allowing financial institutions to understand the factors driving creditworthiness and risk. Talaat et al. [146] showed that XAI also provides a breakdown of credit scores, offering transparency into how individual scores are calculated and the factors influencing them. Moreover, it enables comparative analysis, allowing institutions to compare different models or approaches to credit scoring and risk assessment to determine the most effective strategies. Heng and Subramanian [62] presented that XAI further identifies key risk factors associated with credit decisions, aiding in the identification and mitigation of potential risks. The concept of AI has helped to gain insight into the marketplace, for example, Liu et al. [94, 96] pinpointed that 279.87 million US internet job posts between 2010 and 2022, the demand for statistical talent with expertise in AI increased by an astounding 31 times, resulting in 932 different AI-related employment categories. This insight assists in enhancing the recruitment and employability of job seekers. Additionally, in the work of Fritz-Morgenthal, et al. [48], XAI ensures regulatory compliance by providing explanations and justifications for credit decisions, helping financial institutions adhere to regulatory requirements and ethical standards while making informed lending decisions. Overall, XAI enhances transparency, accountability, and effectiveness in credit scoring and risk assessment processes, enabling financial institutions to make more accurate and fair credit decisions while ensuring regulatory compliance. The banking industry faces challenges in loan acceptance, with rates dropping to 61\u201370% and further decreasing to 50% post-pandemic, leaving many applicants unaware of the reasons for their rejections, which negatively impacts their credit scores. To address this issue, Nallakaruppan et al. [110] proposed an XAI framework integrated with Random Forest algorithms, offering clear explanations for loan decisions, thus enhancing transparency in the lending process. This XAI framework utilizes LIME and SHAP explainers to provide local and global insights on various parameters, significantly improving the accuracy and reliability of loan acceptance predictions.\n\n(iii) Retail and e-commerce: Explainability is gaining increasing significance in retail and e-commerce across various applications, from tailored product suggestions to fraud detection. In the fast-paced world of technology, companies in e-commerce are leveraging AI through applications like recommendation systems, fraud detection, and fake filters to enhance competitiveness. However, a significant challenge remains in understanding and explaining the outcomes produced by AI algorithms, which impacts their trustworthiness and raises ethical and privacy concerns. In e-commerce, where timely decisions are essential, integrating XAI systems is vital. Chaudhary et al. [27] worked systems that are designed to deliver immediate justifications, addressing the shortcomings of existing technologies that often fail to provide comprehensive explanations of the decision-making processes underlying AI-generated results or recommendations. The implementation of XAI not only fosters customer trust and satisfaction but also equips retailers to make informed decisions and streamline their operations [13]. It enables retailers to harmonize personalized experiences with user privacy and ethical considerations. By elucidating the rationale behind individual recommendations or decisions, explainable AI enhances the speed and accuracy of decision-making processes by complementing consumers\u2019 interactions with AI systems. Leveraging customer data, explainable AI facilitates decision-making, leading to improved business outcomes. This transparency and reasoning contribute to reducing abandoned products, increasing order value, and consequently boosting revenue and conversion rates, as noted [78]. One prominent application of AI in e-commerce is the customization of product recommendations, leveraging insights from prior purchases, searches, and browsing behaviors to enhance average order value. Moreover, AI-driven inventory management leverages historical sales data, predicted demand shifts, and potential supply chain issues to optimize inventory levels, unlike traditional methods limited to current stock levels. Additionally, online retailers are adopting chatbots and digital assistants to offer round-the-clock customer support, as highlighted [60].\n\n(iv) Agriculture: Sabrina et al. [134] state that AI explainability plays a vital role in the agriculture industry, helping farmers, agronomists, and policymakers understand and trust AI-driven recommendations and decisions. Explainability in agriculture supports sustainable farming practices, optimal resource utilization, and resilient crop management. It also helps agricultural stakeholders to make informed decisions, improve yields, reduce resource waste, and address ethical and regulatory considerations while fostering trust in AI-driven solutions. The deployment of XAI accounts for the experience and authority of the farmers and comprehension increases the farmer\u2019s trust in AI systems and makes them more effective [64]. Farmers can explore the knowledge acquired by machine learning models trained to identify suitable crops for cultivation through eXplainable AI (XAI) methods and visualizations. With XAI charts, even individuals without expertise in machine learning can comprehend the rationale behind a model\u2019s prediction for a specific crop in a given scenario [23]. Leveraging ML applications like computer vision and time-series analysis, farmers can anticipate the ideal timing for seeding, harvesting, and marketing their crops, thereby enhancing their decision-making capabilities [50]. Effective decision-making processes leading to high and high-quality crop yields involve a comprehensive understanding of weather patterns, along with the use of crop prediction methodologies that consider various soil parameters and atmospheric conditions [41]. XAI plays a transformative role in agriculture by empowering farmers to make informed decisions through transparent insights into crop health, growth patterns, and nutrient requirements derived from data-driven analyses. Moreover, Linheiro, et al. [91] work shows that XAI enhances weather forecasting accuracy by integrating complex weather data with historical patterns, allowing farmers to anticipate and mitigate the impact of adverse weather conditions on crops. In precision agriculture, XAI optimizes resource allocation by offering detailed insights into soil conditions, moisture levels, and crop health at a granular level, facilitating targeted interventions such as irrigation and fertilization. Mana et al. [102], explain that XAI aids in harvest prediction by analyzing historical data, crop maturity levels, and environmental factors to forecast optimal harvest times, maximizing yield, and minimizing waste. Furthermore, XAI contributes to disease and pest detection by leveraging machine learning algorithms to identify early signs of crop diseases and pest infestations, enabling proactive measures such as pest control and crop rotation to protect crop yields and ensure food security. In the study of Chang-Brahim, et al. [26], Marker-assisted selection (MAS) significantly enhances crop breeding by enabling the quick identification of plants with desired traits, with precise phenotyping being essential for its effectiveness. Recent advancements in high-throughput remote phenotyping using unmanned aerial vehicles and machine learning offer efficient alternatives to traditional methods, while AI and explainable AI address the limitations of genome-wide association studies (GWAS) in understanding complex traits like drought tolerance. This review outlines the state-of-the-art methods and processes in MAS, highlighting the integration of explainable AI throughout the breeding workflow from phenotyping to association analyses. Overall, XAI empowers farmers with actionable insights and decision support tools to optimize crop productivity, improve resilience against environmental challenges, and enhance sustainability in agriculture practices.\n\n(v) Education: Explainable Artificial Intelligence plays a crucial role in elucidating the insights uncovered by AI systems, which can unveil novel patterns and discoveries from data, thereby opening up new avenues for research across various disciplines. XAI has the potential to revolutionize the way students perceive, absorb, and comprehend course materials, facilitating enhanced knowledge transfer and optimization of classroom learning experiences. Moreover, XAI can provide valuable insights to educators in learning analytics, highlighting areas where students may require additional support and targeted intervention as well as reshaping the lack of knowledge-sharing system [125, 128]. By promoting transparency and accountability, XAI assists educational institutions and service providers in addressing trust issues associated with AI usage, thereby unlocking its potential benefits [77]. XAI is reshaping education across various facets, starting with adaptive learning, where it tailors instructional content and pace to individual student needs based on comprehensive data analysis, fostering personalized learning experiences and improving student outcomes. In assessment and grading, XAI enhances objectivity and fairness by providing transparent insights into grading criteria and performance metrics, enabling educators to make informed evaluations and provide constructive feedback [67, 69]. Moreover, Lee et al. [89], expressed that XAI empowers educators with learning analytics tools that analyze student data to identify patterns, trends, and areas for improvement, facilitating data-driven decision-making and instructional planning. In resource allocation, XAI optimizes the allocation of educational resources such as personnel, materials, and technology by analyzing student needs, preferences, and performance data, ensuring efficient utilization and equitable access [13]. Also, XAI combined with federated learning (FL) and multimodal fusion eliminates low-quality training models on the server side, minimizing the detrimental effects of low-quality data, data duplication, and inconsistency reasoning without data transmission for human-centric virtual reality [160, 165, 172]. As a rising approach to enhancing trust in AI systems, XAI advocates for the development of methods that provide clear explanations and justifications for the decisions made by AI. Liu, et al. [94, 96] offer an overview of the applications of XAI in education, including examples from empirical studies that address several key questions: What are the primary categories of XAI techniques employed to interpret and clarify AI models? What distinct features characterize the application of XAI in the educational context? How can XAI models assist teachers, students, and researchers in making informed decisions? Additionally, the work explores the challenges and opportunities associated with the use of XAI in education. Additionally, XAI promotes parent and guardian engagement by providing transparent explanations of student progress, learning goals, and educational interventions, fostering collaboration between educators and families to support student success and well-being. Overall, XAI enhances the effectiveness, efficiency, and equity of education by providing transparent insights, personalized support, and data-driven decision-making tools to educators, students, and families alike.\n\n(vi) Transport and Logistics: Explainable Artificial Intelligence is transforming transportation and logistics by offering transparent insights and facilitating data-driven decision-making across diverse sectors. Automobile traffic accidents [166] pose a significant threat to public safety, leading to numerous injuries and fatalities each year. Abdulrashid et al. [1] present a comprehensive design for an explainable artificial intelligence (XAI) artifact that integrates accident data for use by various stakeholders and decision-makers. It introduces responsible and interpretable models along with a taxonomy that categorizes driver-related behaviors linked to different levels of injury severity. In the initial phase, advanced techniques such as Bayesian dynamic conditional imputation for missing data and SHapley Additive exPlanations (SHAP) for risk factor analysis were employed to address data issues and uncover spatiotemporal elements influencing accidents. The final phase includes the classification of crash data into causal factors using aggregate SHAP scores, demonstrating how different risk factors contribute to varying levels of injury severity. The enhanced transparency provided by XAI has the potential to enable traceability, leading to a more precise evaluation of due diligence and liability in the event of accidents, thereby benefiting insurance companies and regulatory bodies. For example, Kuznietsov et al. [85] highlighted the advantages of XAI in the inspection, debugging, and auditing of AI models, which can enhance their robustness and elevate trust levels in autonomous vehicles. XAI plays a crucial role in reinstating the contestability of decisions made by autonomous driving systems, potentially safeguarding users from undue harm and establishing a basis for legal recourse. Mahbooba et al. [100] underscored the necessity of XAI in Intelligent Connected Vehicles (ICV) for reducing vulnerability and detecting intrusions, emphasizing the importance of explanation and interpretability of classification model decisions, as well as effectively managing evolving threats and attack mechanisms. XAI also fosters the development of machine learning algorithms that employ more explainable techniques while maintaining efficient learning performance. Furthermore, it facilitates human comprehension of model actions and decisions, aiding in information extraction and the visualization of results with maximal transparency [115]. In route planning and optimization, XAI analyses historical data, wireless enabled self-powered sensing for logistics in an industry, traffic patterns, and logistical constraints to recommend efficient routes and scheduling, minimizing costs and maximizing delivery efficiency [93]. XAI enhances demand forecasting accuracy by analyzing market trends, customer behaviour, and historical data to predict future demand, enabling companies to optimize inventory management and resource allocation. In freight load optimization, XAI leverages predictive analytics and optimization algorithms to maximize cargo capacity utilization, reducing empty miles and fuel consumption while increasing profitability. In the work El Makhloufi [40], XAI enables real-time tracking of shipments and vehicles, providing stakeholders with visibility into the supply chain and enabling proactive interventions to address delays or disruptions. Furthermore, XAI helps mitigate the environmental impact of transportation by optimizing route planning, load distribution, and vehicle utilization to minimize emissions and carbon footprint, promoting sustainable and eco-friendly practices in the transportation and logistics industry (T. [18, 95]). Overall, XAI empowers transportation and logistics companies with transparent insights and data-driven solutions to optimize operations, improve efficiency, and reduce environmental impact across the supply chain.\n\n(vii) Legal and Regulatory: In legal and regulatory contexts, Explainable AI is applied to clarify and justify AI model decisions, ensuring alignment with legal and ethical standards. XAI aids regulators in understanding complex AI algorithms, making it easier to detect and prevent biases, discrimination, or unfair practices in areas like loan approvals, hiring, and sentencing. By enhancing transparency, XAI fosters accountability, supports audits, and helps organizations meet compliance obligations, such as the GDPR in the EU, which mandates the right to explanations for automated decisions. Fresz et al. [47] paper examines the intersection of law and eXplainable Artificial Intelligence, noting that while the AI Act adopted by the European Parliament in March 2024 is widely discussed, other legal areas remain underexplored. Focusing primarily on European and, to some extent, German law, it also considers international concepts such as fiduciary duties, the General Data Protection Regulation (GDPR), and product safety and liability. By analysing XAI taxonomies, the paper identifies specific requirements for XAI methods in various legal domains, concluding that each area necessitates different properties from XAI. It highlights that the current advancements in XAI do not adequately meet these legal requirements, particularly concerning the correctness (or fidelity) and confidence estimates of XAI methods. This transparency builds trust in AI systems and ensures they adhere to ethical and legal frameworks. Kesari et al. [76], article address the legal need for explainable AI (XAI) in automated decision-making, which often lacks transparency, posing risks to accountability and the public trust which introduces a Legal-XAI framework that categorizes explanations for AI decisions in law, distinguishing between types like global vs. local and comprehensive vs. selective explanations. The authors explore how legal, technical, and behavioral factors influence the choice of explanation method, using credit scoring as a practical example. They advocate for policymakers to specify explanation requirements to align with their objectives, making automated decisions more understandable and democratic. The article bridges gaps between legal and computer science fields, offering a roadmap and software to compare algorithmic explanation methods through empirical field experiments. Mueller et al., (2022), presented that AI is rapidly advancing in healthcare, especially in the field of in vitro diagnostics, but the new European regulation on In Vitro Diagnostic Medical Devices (IVDR) requires AI software in medical devices to be explainable. This requirement challenges the use of machine learning and deep learning in these devices, as they often lack necessary explanatory components. Trustworthy AI is needed to enable healthcare professionals to make informed, accountable decisions, which makes explainable AI essential. Techniques like layer-wise relevance propagation can help visualize which parts of the input data influenced a neural network\u2019s decision. This paper introduces 'causability,' a concept for measuring the quality of AI explanations, which alongside explainability, is crucial for validating the scientific, analytical, and clinical performance of AI-based diagnostic tools. Tax administrations worldwide are increasingly utilizing artificial intelligence (AI) systems for automation. However, this automation significantly impacts the rights of taxpayers who are subject to algorithmic assessments, a situation exacerbated by the complexity and opacity of these AI systems. Ku\u017aniacki et al. [84] article contends that protecting taxpayers\u2019 rights requires the implementation of XAI technologies, which can make the operations and decisions of tax-related AI systems comprehensible to taxpayers, administrative appeal bodies, and the courts. This need arises from constitutional principles governing taxation but remains inadequately addressed by both soft and hard law frameworks, which fail to meet the specific informational needs of the tax sector. To fill this gap, the authors concluded by identifying the technical and legal challenges associated with effectively applying explanation techniques to tax AI, ensuring that automation does not compromise taxpayers\u2019 rights.", "cleaned_text_custom_parser": "Recent Emerging Techniques in Explainable Artificial Intelligence to Enhance the Interpretable and Understanding of AI Models for Human\nAbstract\nRecent advancements in Explainable Artificial Intelligence (XAI) aim to bridge the gap between complex artificial intelligence (AI) models and human understanding, fostering trust and usability in AI systems. However, challenges persist in comprehensively interpreting these models, hindering their widespread adoption. This study addresses these challenges by exploring recently emerging techniques in XAI. The primary problem addressed is the lack of transparency and interpretability in AI models to humanity for institution-wide use, which undermines user trust and inhibits their integration into critical decision-making processes. Through an in-depth review, this study identifies the objectives of enhancing the interpretability of AI models and improving human understanding of their decision-making processes. Various methodological approaches, including post-hoc explanations, model transparency methods, and interactive visualization techniques, are investigated to elucidate AI model behaviours. We further present techniques and methods to make AI models more interpretable and understandable to humans including their strengths and weaknesses to demonstrate promising advancements in model interpretability, facilitating better comprehension of complex AI systems by humans. In addition, we provide the application of XAI in local use cases. Challenges, solutions, and open research directions were highlighted to clarify these compelling XAI utilization challenges. The implications of this research are profound, as enhanced interpretability fosters trust in AI systems across diverse applications, from healthcare to finance. By empowering users to understand and scrutinize AI decisions, these techniques pave the way for more responsible and accountable AI deployment.\nSimilar content being viewed by others\nIntroduction to Explainable AI\nStudy on\u00a0the\u00a0Helpfulness of\u00a0Explainable Artificial Intelligence\nTransformative impact of explainable artificial intelligence: bridging complexity and trust\nExplore related subjects\n1Introduction\nArtificial Intelligence (AI) has played an increasingly significant role in using data collected from diverse sources such as social media, sensors, education, multimedia, and text data, to make necessary decisions that have helped humans [67,68,69]. Explainable Artificial Intelligence (XAI) stems from the increasing integration of AI models in a way that allows humans to understand, interpret, and trust the decisions and outputs generated by AI systems because these systems have become more sophisticated and pervasive. There is a growing need for transparency and interpretability in decision-making processes. With confidence in the deployment of AI by industries and governments, users are seeking enhanced transparency facilitated by XAI systems where the author introduces XAI concepts and provides an overview of areas requiring additional investigation, such as type-2 fuzzy logic systems [58]. The aim guarantees these systems are comprehensible and analyzable by users without specialized expertise. Understanding how AI models arrive at specific conclusions or recommendations is important for building trust and acceptance among users, stakeholders, and society. Ensuring that AI systems are accurate and interpretable helps address concerns related to bias, fairness, and unintended consequences [98]. XAI techniques enable practitioners and regulators to assess the ethical implications of AI decisions, identify potential biases, and mitigate any adverse effects which are particularly relevant in fields such as healthcare, finance, manufacturing, and criminal justice, where the consequences of AI decisions can have significant real-world impact [123].\nExplainable Artificial Intelligence (XAI) lies in its potential to bridge the gap between the complexity of advanced AI models and the need for human understanding and trust. As AI systems continue to play an increasingly pivotal role in critical decision-making processes, it is imperative to enhance their interpretability, making their inner workings more transparent and accessible to users. This not only promotes accountability and ethical use of AI but also addresses concerns related to unintended concerns in empowering individuals, regulators, and stakeholders to comprehend and scrutinize AI models, fostering a responsible and trustworthy integration of artificial intelligence into various aspects of society. In an earlier study by Nott [114], XAI makes AI more transparent and explainable, addressing the \u201cblack box\u201d problem. However, the results achieved do not clearly explain how or why they arrived at those conclusions. Taghikhah et al. [145] and Adadi and Berrada [2] presented \"Cracking open the black box of AI\" refers to the concept of making AI systems more transparent and understandable. AI often operates as a \"black box\" because it can be challenging to interpret how these systems arrive at their decisions or predictions, especially in complex deep learning models and other inner workings of AI systems. Cracking opens the black box of AI, which is essential for building trust in AI systems, ensuring that AI operates ethically, and allowing users to understand and verify AI decisions. Nonetheless, these require much critical attention and development as AI continues to be integrated into various aspects of society [155].\nTo make AI processes more understandable, several key strategies in XAI techniques like model-agnostic interpretability methods can be employed and incorporated to enable users to interpret and comprehend the decisions made by complex AI models [37]. Also, simpler and more interpretable models whenever possible are used instead of highly complex black-box models for transparency to provide clear and concise explanations for AI outputs, including the factors influencing decisions, and enhance user understanding. Furthermore, fostering transparency in the data used to train these models, addressing biases, and regularly auditing and validating AI systems contribute to the overall interpretability and trustworthiness of AI processes because collaborative efforts between AI developers, domain experts, and end-users in the design and deployment stages also play a crucial role in making AI processes more understandable and aligned with human values and expectations [37]. Application of XAI in diverse areas such as healthcare, finance, legal, autonomous vehicles, robotics, etc. enhances the interpretability and understanding of AI models for humans. For instance, Amann et al. [5] utilized XAI to assist clinicians to comprehend and trust complex medical diagnoses and treatment recommendations made by AI systems, ensuring that decisions align with medical expertise. Rane et al. [127] implored XAI in finance to help in explaining algorithmic trading strategies and risk assessment, enabling financial professionals and regulators to make informed decisions and maintain market transparency. In legal and criminal justice systems, XAI assists in explaining the rationale behind AI-generated predictions, ensuring fairness and accountability [90]. More so, Fern\u00e1ndez Llorca, and G\u00f3mez [46] utilized XAI in autonomous vehicles and robotics, to enhance the transparency of decision-making processes, fostering user trust and safety. These applications underscore the importance of XAI in promoting transparency, accountability, and user comprehension across various domains where AI plays a pivotal role.\nDespite the promising potential, Saeed and Omlin [136] reported that XAI faces several challenges in enhancing the interpretability and understanding of AI models for humans, such as the inherent trade-off between model complexity and interpretability, as highly sophisticated models often sacrifice transparency for performance. Also, developing commonly accepted standards for interpretability poses a significant challenge, as different domains may require tailored explanations, and resolving the issue of context-aware explanations is another obstacle, as the interpretability of AI models should account for the specific needs and knowledge levels of diverse user groups. Striking a balance between accuracy and simplicity in explanation generation remains a persistent challenge, especially in complex, high-dimensional datasets, Kaushik [75]. Furthermore, Chamola, et al. [25] reviewed that ensuring that XAI methods effectively handle real-world scenarios, including unforeseen situations and adversarial attacks, is essential for their practical applicability, and overcoming these challenges is imperative for the widespread adoption of XAI, ensuring that AI systems are not only accurate but also transparent, trustworthy, and accessible to a broad range of users.\nProgressively, various studies have attempted to address these peculiar challenges. For instance, Lythe et al. [99] pinpointed that interpretability goes beyond explainability, aiming to make AI models more accessible to individuals, particularly domain experts unfamiliar with intricate AI techniques, by demystifying complex model architectures. Felzmann et al. [45] discussed that transparency in AI involves disclosing information about data, algorithms, and processes, enhancing visibility throughout the AI development and deployment stages. Additionally, ongoing research focuses on developing various explainable AI techniques, such as feature attribution methods, rule-based models, and attention mechanisms that unveil influential aspects of input data in model decisions. However, to our knowledge, none of these studies provides a comprehensive approach in solving the inherent issues beclouding the deployment of AI. Ehsan et al. [39] qualitatively explain how an individual's AI background can shape their interpretations, highlighting differences through the perspectives of appropriation and cognitive heuristics. The study finds that both groups exhibited undue trust in numerical data for different reasons and appreciated various explanations beyond their intended design. These findings, which have significant implications for XAI, reveal that AI-generated explanations can inadvertently lead to negative consequences and potentially harmful manipulation of trust. The authors propose design interventions to address these issues.\nConversely, this paper addresses a significant research gap in the field of artificial intelligence (AI) because it primarily revolves around the pressing need for AI models to be more transparent and interpretable to humans. Therefore, the contributions of this paper to the current body of knowledge are as follows:\nWe provided a comprehensive understanding of the recent emerging techniques in XAI that address an in-depth dimension of explanatory;\nWe investigated and evaluated the efficacy of specific techniques in XAI, such as sensitivity analysis, rule-based models, documentation and transparency methods, prototype-based explanations, and visualization techniques;\nProvides the strengths, weaknesses, and features of AI components, including machine learning, natural language processing, computer vision, knowledge representation and reasoning, robotics, and expert systems;\nWe explore an in-depth utilization of XAI techniques in diverse local use cases spanning healthcare diagnostics, financial risk assessment, retail and e-commerce, agriculture, education, transport, and logistics;\nWe further identified and analyzed the challenges associated with deploying XAI techniques in local use cases across different domains;\nThe rest of the paper is organized into the following sections: Sect.2delves into the explanatory depth of XAI. Section2.1outlines techniques and methods to enhance the interpretability and understandability of AI models for humans, while Sect.2.2discusses the understandability of AI components in detail. Furthermore, Sect.3gave an in-depth discussion on the deployment of AI explainable in local use cases while Sect.4covered challenges and solutions of AI explainable in local use cases. Finally, Sect.5handled areas on open research direction.\n2Explanatory Depth of Explainable AI\nIn the context of XAI, explanatory depth refers to the extent of detail and clarity in the explanations offered by AI systems. Some works have provided explanatory depth on XAI; Chromik et al. [30] stated that explanatory depth shows how deeply an AI system can explain its decisions, going beyond seeming or shallow explanations and the concept of explanatory depth is vital because not all AI explanations are equally informative or useful. Some AI systems may provide high-level, imprecise explanations that do not help users understand the underlying reasoning, while others can offer detailed, insightful explanations that enhance user\u2019s trust and comprehension [61]. So, Hulsen [65] in his work stated that XAI gives the richness and detail of the explanations provided by AI systems in making AI more understandable and trustworthy, and achieving it involves various techniques and considerations to ensure that the explanations are meaningful and helpful to users. There are various methods used to achieve explanatory depth in XAI and these techniques include, Antoniadi et al. [8] reviewed that feature importance that is the most influential feature in the decision-making process is pointed out, Eberhard, [38] extracted that visualization using visual aids shows the decision process and feature interactions, and Mittelstadt et al. [106] explained that counterfactual explanations by showing what changes in input data would have resulted in a different decision and local verse global explanations that offer insight into how a specific decision was made (local) as well as how the overall model behaves (global). A recent study by Del Ser et al. [34], introduced a framework for generating counterfactual explanations, formulating the task as a multi-objective optimization problem that balances three goals: plausibility, change intensity, and adversarial effectiveness which uses generative adversarial networks (GANs) to model input distribution, alongside a solver for multi-objective counterfactual discovery, the framework addresses the envisaged peculiar problems. The study demonstrates across six classification tasks using image and 3D data confirm evidence of trade-offs between objectives and consistency of the generated explanations with human understanding, and the framework's ability to reveal concept-based biases and misrepresented attributes within the model's input domain. Therefore, XAI techniques, including feature importance, visualization, and counterfactual explanations, enhance explanatory depth by enabling users to comprehend and trust the decisions made by AI systems.\nPost-hoc explanation is an external method used after a model is trained to clarify its predictions. For instance, Retzlaff et al. [132] introduced a decision tree that helps navigate key trade-offs in various XAI methods, guiding users in selecting from six popular XAI tools. Post-hoc methods are beneficial for explaining complex models that are not naturally interpretable, such as deep neural networks or ensemble models. Techniques commonly used for post-hoc explanation include feature importance scores, SHapley Additive Explanations (SHAP), and (Local Interpretable Model-agnostic Explanations (LIME), enabling complex models to offer valuable insights into their decision-making processes. Through expert interviews, post-hoc XAI methods will be evaluated for their applicability in real-world contexts. The goal is to enable data scientists and policymakers to choose XAI methods that not only clarify the decision-making process but also enhance user comprehension and interpretation, ultimately fostering the practical adoption of XAI. Vale et al. [150] alluded to the insights from\u00a0post-hoc explainability methods used to help regulate black-box machine learning. Some of the post-hoc explanations include in healthcare, SHAP helps identify how patient features (such as age, symptoms, or lab results) influence a diagnosis or treatment recommendation, building trust in AI-driven decisions. LIME, on the other hand, explains individual predictions, clarifying why specific diagnoses or risk scores were assigned, which is especially useful for case-specific interpretations. In finance, SHAP clarifies how factors like credit history, income, and debt impact loan approvals or risk assessments, supporting regulatory compliance and customer trust. LIME offers case-specific explanations for individual loan or credit decisions, helping both analysts and customers understand specific outcomes. In agriculture, SHAP explains predictions on crop yield, disease risk, and soil quality by highlighting the impact of factors like weather, soil pH, and fertilizer use, supporting data-driven crop management while LIME offers specific explanations for cases such as high disease risk scores, guiding targeted interventions. In education, SHAP clarifies how factors like attendance, grades, and engagement affect predictions of student performance or dropout risk, helping educators target areas for intervention. LIME provides case-specific insights, allowing educators to understand why particular students are identified as high-risk and address their unique needs. In retail, SHAP highlights key customer features, like purchase history and demographics, that influence behavior predictions, aiding in personalization strategies. LIME provides specific explanations for individual recommendations, allowing retailers to refine algorithms and enhance customer experience. In transport, SHAP identifies factors affecting delivery times, maintenance needs, or fuel efficiency\u2014such as traffic, vehicle age, or route length\u2014helping to boost efficiency. LIME offers explanations for specific delays or routing choices, enabling managers to understand and adjust schedules as needed.\nZacharias et al. [169] explained that the feature importance technique seeks to improve explanatory depth by pinpointing the most significant features influencing an AI model's decisions. By determining which features most affect model predictions, users gain insights into the factors driving the AI system's behaviour. It allows users to prioritize their attention and understand the underlying logic of the model so that users can focus on relevant factors and comprehend why certain decisions are made, thereby increasing the transparency and interpretability of the AI system. Visualization techniques play a crucial role in achieving explanatory depth by providing intuitive representations of AI model behaviour by transforming complex model outputs into visually interpretable formats, such as charts, graphs, or heatmaps, users can better understand the patterns and relationships captured by the AI system Minh et al. [105]. Visualization techniques enable users to explore the inner workings of the model, identify trends, and detect anomalies more effectively. Furthermore, interactive visualizations allow users to manipulate data inputs and observe real-time changes in model predictions, facilitating a deeper understanding of how the AI system responds to different scenarios. In the work of Stepin et al. [141] counterfactual explanations offer another avenue for achieving explanatory depth in XAI by providing insights into \u201cwhat-if\u201d scenarios by generating hypothetical instances where model predictions change, users can gain a better understanding of the decision boundaries and causal relationships within the AI system which help users explore alternative outcomes and understand how changes in input variables influence model predictions. The goals of Explainable AI are presented in Fig.1.\nFigure1illustrates that the primary objective of Explainable Artificial Intelligence (XAI) is to improve the comprehension and acceptance of AI systems. This is achieved by integrating essential principles such as transparency, reliability, causality, usability, privacy, trust, and fairness. Transparency ensures that the internal mechanisms of AI models are clear and understandable, thereby promoting trust and accountability. Reliability focuses on consistently producing accurate and dependable results, contributing to the overall credibility of AI systems. Causality aims to uncover the cause-and-effect relationships within AI decisions, providing deeper insights into model behaviour. Usability emphasizes making AI systems accessible and user-friendly, catering to diverse stakeholders. Privacy concerns involve safeguarding sensitive information and ensuring responsible data handling practices. Trust is built through transparent processes, reliable outcomes, and ethical considerations. Fairness addresses the need for unbiased and equitable AI decisions, mitigating the impact of algorithmic biases, and thereby promoting ethical and responsible AI practices across various domains.\n2.1Techniques and Methods to Make AI Models More Interpretable and Understandable to Humans\nWeber et al. [163,164] highlighted the importance of making AI models more interpretable and understandable to humans for building trust and ensuring responsible deployment of AI systems. AI practitioners can employ certain techniques and methods to enhance model interpretability and understandability, fostering responsible AI deployment and increasing user trust. Table1presents some of these techniques and methods.\n2.2Understandability of Artificial Intelligence Components\nArtificial Intelligence (AI) is a diverse and multidisciplinary field encompassing numerous elements, methodologies, and specialized areas. These interconnected AI components often collaborate within AI systems to tackle intricate challenges and execute tasks [70]. The realm of AI research and development continually progresses, yielding novel techniques and breakthroughs across its various domains. In studies by Arrieta et al. [10] and Montavon et al. [108], \"understandability\" refers to both the extent to which a human can comprehend a model\u2019s decision and the model\u2019s capacity to elucidate its function to a human. AI comprises several key components that synergize to empower machines to undertake tasks typically requiring human intelligence. These AI components undertake diverse tasks, spanning from natural language comprehension and generation to image object recognition and autonomous decision-making. By harnessing these capabilities, AI systems enhance human intelligence, automate repetitive tasks, and tackle complex challenges across diverse fields. Table2elucidates some of these key components.\n3Application of AI Explainable in Local Use Cases\nAI explainability has many local use cases across different industries and applications [15]. The local use cases include healthcare, finance, retail and e-commerce, agriculture, education, transport and logistics, and legal and regulatory. These are succinctly presented in the Fig.2and discussed as follows:\n(i) Healthcare: Explainable Artificial Intelligence (XAI) plays a crucial role in disease diagnosis, AI models used in medical diagnosis can explain their decisions, helping doctors understand the reasoning behind a diagnosis. Ensuring explainability in AI for disease diagnosis is crucial for fostering trust, transparency, and acceptance of AI-based diagnostic systems within the healthcare sector. As highlighted in study by Asan et al. [11], when patients and healthcare providers comprehend the reasoning behind AI-generated diagnoses, they can make more informed decisions and place greater trust in the recommendations. Incorporating explainability into AI disease diagnosis not only assists healthcare professionals but also addresses concerns regarding safety, accountability, and ethical utilization of AI in healthcare [21]. It is imperative to strike a balance between model accuracy and interpretability to establish trust in the AI system and enable its effective integration into medical practice. According to Zhang et al. [170], eXplainable AI in healthcare focuses on several critical aspects, including ensuring transparent data and features, thereby enabling healthcare providers to comprehend the data inputs and variables influencing the diagnostic process. Secondly, XAI provides insights into feature importance and contribution, enabling clinicians to identify critical factors contributing to a diagnosis. Additionally, it offers visualization tools and graphs to represent complex data relationships visually, aiding in comprehension and decision-making. Moreover, XAI enhances model interpretability, allowing healthcare providers to understand how a diagnostic model arrives at its conclusions. Furthermore, Schoonderwoerd et al. [140] explained that XAI facilitates natural language explanations, which help communicate diagnostic results and treatment recommendations in a comprehensible manner to both clinicians and patients. XAI also supports treatment recommendations, comparative analysis between different diagnostic approaches, and patient education, thereby empowering healthcare professionals and patients with actionable insights and knowledge for informed decision-making in disease diagnosis and management. For instance, a recent study by Zhu [171] utilized computational intelligence with AI to cclassify brain images from MRI or CT scans, demonstrating the impact of dosage ratios on health over treatment duration, and diagnosing memory impairment in psychoactive substance users. The shift towards digital pathology allows pathologists to utilize AI-based computer programs for analyzing whole slide images, but many top-performing algorithms operate as \"black boxes,\" leaving even developers uncertain about their decision-making processes [120]. This review emphasizes the necessity of explainability in digital pathology to prevent mistakes and ensure patient safety, introducing various explainable AI techniques that enhance the transparency of these algorithms. Furthermore, it advocates for the development of novel user interfaces that promote contextual understanding and interactive exploration, enabling medical experts to engage with AI results more effectively and maintain a critical human presence in the decision-making process.\n(ii) Finance: Weber et al. [163,164], the explainability of AI in finance, particularly in credit scoring and risk assessment, is crucial for several reasons, including regulatory compliance, transparency, and trust-building with both lenders and borrowers. In credit scoring, AI models can explain why an applicant\u2019s credit score is calculated in a certain way, offering transparency to applicants and regulators while in risk assessment, explainable AI models can provide insights into the factors influencing risk assessment in financial investments. Explainability in AI for credit scoring and risk assessment helps lenders make responsible lending decisions and empowers borrowers to understand and improve their financial profiles [62]. Moreover, it assists regulatory bodies in ensuring that AI-driven credit decisions comply with legal and ethical standards, promoting fairness and accountability in the financial industry. Matcov [103] explained that XAI is instrumental in credit scoring and risk assessment within the finance sector, operating across several key dimensions like identifying feature importance, and allowing financial institutions to understand the factors driving creditworthiness and risk. Talaat et al. [146] showed that XAI also provides a breakdown of credit scores, offering transparency into how individual scores are calculated and the factors influencing them. Moreover, it enables comparative analysis, allowing institutions to compare different models or approaches to credit scoring and risk assessment to determine the most effective strategies. Heng and Subramanian [62] presented that XAI further identifies key risk factors associated with credit decisions, aiding in the identification and mitigation of potential risks. The concept of AI has helped to gain insight into the marketplace, for example, Liu et al. [94,96] pinpointed that 279.87 million US internet job posts between 2010 and 2022, the demand for statistical talent with expertise in AI increased by an astounding 31 times, resulting in 932 different AI-related employment categories. This insight assists in enhancing the recruitment and employability of job seekers. Additionally, in the work of Fritz-Morgenthal, et al. [48], XAI ensures regulatory compliance by providing explanations and justifications for credit decisions, helping financial institutions adhere to regulatory requirements and ethical standards while making informed lending decisions. Overall, XAI enhances transparency, accountability, and effectiveness in credit scoring and risk assessment processes, enabling financial institutions to make more accurate and fair credit decisions while ensuring regulatory compliance. The banking industry faces challenges in loan acceptance, with rates dropping to 61\u201370% and further decreasing to 50% post-pandemic, leaving many applicants unaware of the reasons for their rejections, which negatively impacts their credit scores. To address this issue, Nallakaruppan et al. [110] proposed an XAI framework integrated with Random Forest algorithms, offering clear explanations for loan decisions, thus enhancing transparency in the lending process. This XAI framework utilizes LIME and SHAP explainers to provide local and global insights on various parameters, significantly improving the accuracy and reliability of loan acceptance predictions.\n(iii) Retail and e-commerce: Explainability is gaining increasing significance in retail and e-commerce across various applications, from tailored product suggestions to fraud detection. In the fast-paced world of technology, companies in e-commerce are leveraging AI through applications like recommendation systems, fraud detection, and fake filters to enhance competitiveness. However, a significant challenge remains in understanding and explaining the outcomes produced by AI algorithms, which impacts their trustworthiness and raises ethical and privacy concerns. In e-commerce, where timely decisions are essential, integrating XAI systems is vital. Chaudhary et al. [27] worked systems that are designed to deliver immediate justifications, addressing the shortcomings of existing technologies that often fail to provide comprehensive explanations of the decision-making processes underlying AI-generated results or recommendations. The implementation of XAI not only fosters customer trust and satisfaction but also equips retailers to make informed decisions and streamline their operations [13]. It enables retailers to harmonize personalized experiences with user privacy and ethical considerations. By elucidating the rationale behind individual recommendations or decisions, explainable AI enhances the speed and accuracy of decision-making processes by complementing consumers\u2019 interactions with AI systems. Leveraging customer data, explainable AI facilitates decision-making, leading to improved business outcomes. This transparency and reasoning contribute to reducing abandoned products, increasing order value, and consequently boosting revenue and conversion rates, as noted [78]. One prominent application of AI in e-commerce is the customization of product recommendations, leveraging insights from prior purchases, searches, and browsing behaviors to enhance average order value. Moreover, AI-driven inventory management leverages historical sales data, predicted demand shifts, and potential supply chain issues to optimize inventory levels, unlike traditional methods limited to current stock levels. Additionally, online retailers are adopting chatbots and digital assistants to offer round-the-clock customer support, as highlighted [60].\n(iv) Agriculture: Sabrina et al. [134] state that AI explainability plays a vital role in the agriculture industry, helping farmers, agronomists, and policymakers understand and trust AI-driven recommendations and decisions. Explainability in agriculture supports sustainable farming practices, optimal resource utilization, and resilient crop management. It also helps agricultural stakeholders to make informed decisions, improve yields, reduce resource waste, and address ethical and regulatory considerations while fostering trust in AI-driven solutions. The deployment of XAI accounts for the experience and authority of the farmers and comprehension increases the farmer\u2019s trust in AI systems and makes them more effective [64]. Farmers can explore the knowledge acquired by machine learning models trained to identify suitable crops for cultivation through eXplainable AI (XAI) methods and visualizations. With XAI charts, even individuals without expertise in machine learning can comprehend the rationale behind a model\u2019s prediction for a specific crop in a given scenario [23]. Leveraging ML applications like computer vision and time-series analysis, farmers can anticipate the ideal timing for seeding, harvesting, and marketing their crops, thereby enhancing their decision-making capabilities [50]. Effective decision-making processes leading to high and high-quality crop yields involve a comprehensive understanding of weather patterns, along with the use of crop prediction methodologies that consider various soil parameters and atmospheric conditions [41]. XAI plays a transformative role in agriculture by empowering farmers to make informed decisions through transparent insights into crop health, growth patterns, and nutrient requirements derived from data-driven analyses. Moreover, Linheiro, et al. [91] work shows that XAI enhances weather forecasting accuracy by integrating complex weather data with historical patterns, allowing farmers to anticipate and mitigate the impact of adverse weather conditions on crops. In precision agriculture, XAI optimizes resource allocation by offering detailed insights into soil conditions, moisture levels, and crop health at a granular level, facilitating targeted interventions such as irrigation and fertilization. Mana et al. [102], explain that XAI aids in harvest prediction by analyzing historical data, crop maturity levels, and environmental factors to forecast optimal harvest times, maximizing yield, and minimizing waste. Furthermore, XAI contributes to disease and pest detection by leveraging machine learning algorithms to identify early signs of crop diseases and pest infestations, enabling proactive measures such as pest control and crop rotation to protect crop yields and ensure food security. In the study of Chang-Brahim, et al. [26], Marker-assisted selection (MAS) significantly enhances crop breeding by enabling the quick identification of plants with desired traits, with precise phenotyping being essential for its effectiveness. Recent advancements in high-throughput remote phenotyping using unmanned aerial vehicles and machine learning offer efficient alternatives to traditional methods, while AI and explainable AI address the limitations of genome-wide association studies (GWAS) in understanding complex traits like drought tolerance. This review outlines the state-of-the-art methods and processes in MAS, highlighting the integration of explainable AI throughout the breeding workflow from phenotyping to association analyses. Overall, XAI empowers farmers with actionable insights and decision support tools to optimize crop productivity, improve resilience against environmental challenges, and enhance sustainability in agriculture practices.\n(v) Education: Explainable Artificial Intelligence plays a crucial role in elucidating the insights uncovered by AI systems, which can unveil novel patterns and discoveries from data, thereby opening up new avenues for research across various disciplines. XAI has the potential to revolutionize the way students perceive, absorb, and comprehend course materials, facilitating enhanced knowledge transfer and optimization of classroom learning experiences. Moreover, XAI can provide valuable insights to educators in learning analytics, highlighting areas where students may require additional support and targeted intervention as well as reshaping the lack of knowledge-sharing system [125,128]. By promoting transparency and accountability, XAI assists educational institutions and service providers in addressing trust issues associated with AI usage, thereby unlocking its potential benefits [77]. XAI is reshaping education across various facets, starting with adaptive learning, where it tailors instructional content and pace to individual student needs based on comprehensive data analysis, fostering personalized learning experiences and improving student outcomes. In assessment and grading, XAI enhances objectivity and fairness by providing transparent insights into grading criteria and performance metrics, enabling educators to make informed evaluations and provide constructive feedback [67,69]. Moreover, Lee et al. [89], expressed that XAI empowers educators with learning analytics tools that analyze student data to identify patterns, trends, and areas for improvement, facilitating data-driven decision-making and instructional planning. In resource allocation, XAI optimizes the allocation of educational resources such as personnel, materials, and technology by analyzing student needs, preferences, and performance data, ensuring efficient utilization and equitable access [13]. Also, XAI combined with federated learning (FL) and multimodal fusion eliminates low-quality training models on the server side, minimizing the detrimental effects of low-quality data, data duplication, and inconsistency reasoning without data transmission for human-centric virtual reality [160,165,172]. As a rising approach to enhancing trust in AI systems, XAI advocates for the development of methods that provide clear explanations and justifications for the decisions made by AI. Liu, et al. [94,96] offer an overview of the applications of XAI in education, including examples from empirical studies that address several key questions: What are the primary categories of XAI techniques employed to interpret and clarify AI models? What distinct features characterize the application of XAI in the educational context? How can XAI models assist teachers, students, and researchers in making informed decisions? Additionally, the work explores the challenges and opportunities associated with the use of XAI in education. Additionally, XAI promotes parent and guardian engagement by providing transparent explanations of student progress, learning goals, and educational interventions, fostering collaboration between educators and families to support student success and well-being. Overall, XAI enhances the effectiveness, efficiency, and equity of education by providing transparent insights, personalized support, and data-driven decision-making tools to educators, students, and families alike.\n(vi) Transport and Logistics: Explainable Artificial Intelligence is transforming transportation and logistics by offering transparent insights and facilitating data-driven decision-making across diverse sectors. Automobile traffic accidents [166] pose a significant threat to public safety, leading to numerous injuries and fatalities each year. Abdulrashid et al. [1] present a comprehensive design for an explainable artificial intelligence (XAI) artifact that integrates accident data for use by various stakeholders and decision-makers. It introduces responsible and interpretable models along with a taxonomy that categorizes driver-related behaviors linked to different levels of injury severity. In the initial phase, advanced techniques such as Bayesian dynamic conditional imputation for missing data and SHapley Additive exPlanations (SHAP) for risk factor analysis were employed to address data issues and uncover spatiotemporal elements influencing accidents. The final phase includes the classification of crash data into causal factors using aggregate SHAP scores, demonstrating how different risk factors contribute to varying levels of injury severity. The enhanced transparency provided by XAI has the potential to enable traceability, leading to a more precise evaluation of due diligence and liability in the event of accidents, thereby benefiting insurance companies and regulatory bodies. For example, Kuznietsov et al. [85] highlighted the advantages of XAI in the inspection, debugging, and auditing of AI models, which can enhance their robustness and elevate trust levels in autonomous vehicles. XAI plays a crucial role in reinstating the contestability of decisions made by autonomous driving systems, potentially safeguarding users from undue harm and establishing a basis for legal recourse. Mahbooba et al. [100] underscored the necessity of XAI in Intelligent Connected Vehicles (ICV) for reducing vulnerability and detecting intrusions, emphasizing the importance of explanation and interpretability of classification model decisions, as well as effectively managing evolving threats and attack mechanisms. XAI also fosters the development of machine learning algorithms that employ more explainable techniques while maintaining efficient learning performance. Furthermore, it facilitates human comprehension of model actions and decisions, aiding in information extraction and the visualization of results with maximal transparency [115]. In route planning and optimization, XAI analyses historical data, wireless enabled self-powered sensing for logistics in an industry, traffic patterns, and logistical constraints to recommend efficient routes and scheduling, minimizing costs and maximizing delivery efficiency [93]. XAI enhances demand forecasting accuracy by analyzing market trends, customer behaviour, and historical data to predict future demand, enabling companies to optimize inventory management and resource allocation. In freight load optimization, XAI leverages predictive analytics and optimization algorithms to maximize cargo capacity utilization, reducing empty miles and fuel consumption while increasing profitability. In the work El Makhloufi [40], XAI enables real-time tracking of shipments and vehicles, providing stakeholders with visibility into the supply chain and enabling proactive interventions to address delays or disruptions. Furthermore, XAI helps mitigate the environmental impact of transportation by optimizing route planning, load distribution, and vehicle utilization to minimize emissions and carbon footprint, promoting sustainable and eco-friendly practices in the transportation and logistics industry (T. [18,95]). Overall, XAI empowers transportation and logistics companies with transparent insights and data-driven solutions to optimize operations, improve efficiency, and reduce environmental impact across the supply chain.\n(vii) Legal and Regulatory: In legal and regulatory contexts, Explainable AI is applied to clarify and justify AI model decisions, ensuring alignment with legal and ethical standards. XAI aids regulators in understanding complex AI algorithms, making it easier to detect and prevent biases, discrimination, or unfair practices in areas like loan approvals, hiring, and sentencing. By enhancing transparency, XAI fosters accountability, supports audits, and helps organizations meet compliance obligations, such as the GDPR in the EU, which mandates the right to explanations for automated decisions. Fresz et al. [47] paper examines the intersection of law and eXplainable Artificial Intelligence, noting that while the AI Act adopted by the European Parliament in March 2024 is widely discussed, other legal areas remain underexplored. Focusing primarily on European and, to some extent, German law, it also considers international concepts such as fiduciary duties, the General Data Protection Regulation (GDPR), and product safety and liability. By analysing XAI taxonomies, the paper identifies specific requirements for XAI methods in various legal domains, concluding that each area necessitates different properties from XAI. It highlights that the current advancements in XAI do not adequately meet these legal requirements, particularly concerning the correctness (or fidelity) and confidence estimates of XAI methods. This transparency builds trust in AI systems and ensures they adhere to ethical and legal frameworks. Kesari et al. [76], article address the legal need for explainable AI (XAI) in automated decision-making, which often lacks transparency, posing risks to accountability and the public trust which introduces a Legal-XAI framework that categorizes explanations for AI decisions in law, distinguishing between types like global vs. local and comprehensive vs. selective explanations. The authors explore how legal, technical, and behavioral factors influence the choice of explanation method, using credit scoring as a practical example. They advocate for policymakers to specify explanation requirements to align with their objectives, making automated decisions more understandable and democratic. The article bridges gaps between legal and computer science fields, offering a roadmap and software to compare algorithmic explanation methods through empirical field experiments. Mueller et al., (2022), presented that AI is rapidly advancing in healthcare, especially in the field of in vitro diagnostics, but the new European regulation on In Vitro Diagnostic Medical Devices (IVDR) requires AI software in medical devices to be explainable. This requirement challenges the use of machine learning and deep learning in these devices, as they often lack necessary explanatory components. Trustworthy AI is needed to enable healthcare professionals to make informed, accountable decisions, which makes explainable AI essential. Techniques like layer-wise relevance propagation can help visualize which parts of the input data influenced a neural network\u2019s decision. This paper introduces \"causability,\" a concept for measuring the quality of AI explanations, which alongside explainability, is crucial for validating the scientific, analytical, and clinical performance of AI-based diagnostic tools. Tax administrations worldwide are increasingly utilizing artificial intelligence (AI) systems for automation. However, this automation significantly impacts the rights of taxpayers who are subject to algorithmic assessments, a situation exacerbated by the complexity and opacity of these AI systems. Ku\u017aniacki et al. [84] article contends that protecting taxpayers\u2019 rights requires the implementation of XAI technologies, which can make the operations and decisions of tax-related AI systems comprehensible to taxpayers, administrative appeal bodies, and the courts. This need arises from constitutional principles governing taxation but remains inadequately addressed by both soft and hard law frameworks, which fail to meet the specific informational needs of the tax sector. To fill this gap, the authors concluded by identifying the technical and legal challenges associated with effectively applying explanation techniques to tax AI, ensuring that automation does not compromise taxpayers\u2019 rights.\n4Challenges and Solutions of AI Explainable in Local Use Cases\nAs Artificial Intelligence (AI) continues to permeate various sectors and industries, the demand for transparency and interpretability in AI systems has become increasingly pronounced. In local use cases, where AI is applied within specific contexts such as healthcare, finance, retail, agriculture, education, transport, and logistics, ensuring explainability presents unique challenges and opportunities. Work by D\u00edaz-Rodr\u00edguez, et al. [36] discussed the challenges and solutions associated with AI explainability in local use cases, highlighting the importance of understanding how AI systems arrive at their decisions to foster trust, accountability, and ethical AI deployment discussed in this section. By addressing these challenges with tailored solutions, stakeholders can harness the full potential of AI while mitigating risks and ensuring responsible and impactful AI applications in diverse local contexts as shown in Table3.\n5Open Research Directions\n(i) Integration of Human Feedback: One promising research direction involves the integration of human feedback into explainable AI systems to enhance interpretability and understanding. Saeed and Omlin [136] extracted that current XAI techniques aim to provide transparent explanations for AI decisions, incorporating user feedback can further refine these explanations and improve their relevance and usefulness. Research in this area could explore methods for soliciting and incorporating feedback from end-users, domain experts, and stakeholders to iteratively improve the interpretability of AI models. This could involve developing interactive XAI interfaces that allow users to query the system, provide feedback on generated explanations, and contribute domain-specific knowledge to enhance model interpretability.\n(ii) Explainability in Deep Learning Models: Another crucial area of research involves the advancement of explainability methods tailored for deep learning models [166], which often function as opaque entities due to their intricate architectures and high-dimensional feature representations. While recent progress in eXplainable AI (XAI) has made significant strides in interpreting simpler models like decision trees or linear models, deep learning models present distinct challenges for explainability. Future research endeavors could concentrate on devising innovative techniques that offer transparent explanations for deep learning models, including attention mechanisms, convolutional neural networks (CNNs), and recurrent neural networks (RNNs) [66]. This may entail the exploration of approaches for visualizing and interpreting the internal representations of deep learning models, extracting pertinent features, and elucidating the decision-making processes of intricate neural networks.\n(iii) Contextual Explainability: Furthermore, research into contextual explainability is gaining traction, especially for local use cases where AI systems function within specific domains or environments. Contextual explainability involves AI systems offering explanations that are relevant and tailored to the specific context or user preferences. This requires considering factors such as cultural norms, regulatory requirements, and user expectations when generating explanations for AI decisions [33]. Nagahisarchoghaei et al. [109] suggested that future research could focus on creating context-aware XAI techniques that adapt explanations to the distinct characteristics of various local use cases, ensuring that AI systems deliver interpretable and actionable insights that are meaningful and relevant to users.\n(iv) Ethical and Social Implications: Future research in XAI should address the ethical and social implications of explainable AI systems, particularly in local use cases where AI applications can significantly impact society. This includes tackling issues such as fairness, transparency, accountability, and privacy in AI decision-making processes. Researchers could investigate methods for detecting and mitigating biases in XAI systems to ensure explanations are accurate, unbiased, and trustworthy. Additionally, it is important to examine the broader societal implications of XAI, including its effects on job displacement, human-AI collaboration, and the distribution of power and resources in society [139]. By considering these ethical and social factors, researchers can ensure that XAI technologies are developed and deployed responsibly, with careful consideration of their potential impacts on individuals, communities, and society at large.\n6Conclusion\nThe recent advancements in Explainable Artificial Intelligence (XAI) represent a crucial step toward bridging the gap between complex AI models and human understanding. By improving explanatory depth, XAI techniques provide users with insights into AI decision-making processes, thereby enhancing transparency and trust. As AI continues to evolve and integrate into various societal sectors, the development of effective explanation methods remains a priority in AI research and development. These explanations aim to make AI system behavior transparent and interpretable to humans, fostering trust, accountability, and informed decision-making across various applications. As demonstrated in this work, developers, operators, and users generally expect XAI to clarify key questions, enabling people to fully understand and trust AI system decisions. This trust will allow users to have confidence in the models and encompass all XAI components related to transparency, causality, bias, fairness, and safety. However, the lack of explainability and trust currently limits the full utilization of AI systems.\nFurthermore, the implementation of XAI in local use cases has great potential to improve decision-making processes in various fields such as healthcare, finance, and law. However, successful utilization requires careful consideration of local data characteristics, regulatory requirements, and user preferences. As XAI continues to evolve, several key areas of future research will shape its development. These include enhancing model transparency by devising methods to make complex AI models, like deep learning models, more transparent and interpretable without significantly compromising their performance. This research offers practical implications, for instance, facilitating better decision-making processes in industries by providing transparent and interpretable insights into AI models. Also, implementing emerging techniques in XAI can foster trust and acceptance among users, as they gain a deeper understanding of AI model behaviour and reasoning, leading to increased confidence in the technology. Furthermore, the study emphasizes the importance of collaboration between computer scientists, ethicists, psychologists, sociologists, and other disciplines, highlighting the need for a holistic approach to address the multifaceted challenges of XAI effectively.\nData and Material Availability\nCode Availability\nReferences\nAbdulrashid I, Farahani RZ, Mammadov S, Khalafalla M, Chiang WC (2024) Explainable artificial intelligence in transport logistics: risk analysis for road accidents. Transp Res Part E: Logist Transp Rev 186:103563\nAdadi A, Berrada M (2018) Peeking inside the black box: a survey on explainable artificial intelligence (XAI). IEEE Access 6:52138\u201352160\nAlicioglu G, Sun B (2022) A survey of visual analytics for explainable artificial intelligence methods. Comput Graph 102:502\u2013520\nAloisi C (2023) The future of standardised assessment: validity and trust in algorithms for assessment and scoring. Eur J Educ 58(1):98\u2013110\nAmann J, Blasimme A, Vayena E, Frey D, Madai VI,Precise4Q Consortium (2020) Explainability for artificial intelligence in healthcare: a multidisciplinary perspective. BMC Med Inf Decis Mak, 20, pp 1\u20139\nAndreu-Perez J, Deligianni F, Ravi D, Yang GZ (2018) Artificial Intelligence and Robotics. arXiv preprintarXiv:1803.10813.\nAnica-Popa I, Anica-Popa L, R\u0103dulescu C, Vr\u00eencianu M (2021) The integration of artificial intelligence in retail: benefits, challenges and a dedicated conceptual framework. Amfiteatru Economic 23(56):120\u2013136\nAntoniadi AM, Du Y, Guendouz Y, Wei L, Mazo C, Becker BA, Mooney C (2021) Current challenges and future opportunities for XAI in machine learning-based clinical decision support systems: a systematic review. Appl Sci 11(11):5088\nAraujo H, Mousavi MR, Varshosaz M (2023) Testing, validation, and verification of robotic and autonomous systems: a systematic review. ACM Transact Softw Eng Methodol 32(2):1\u201361\nArrieta AB, D\u00edaz-Rodr\u00edguez N, Del Ser J, Bennetot A, Tabik S, Barbado A, Herrera F (2020) Explainable Artificial Intelligence (XAI): concepts, taxonomies, opportunities, and challenges toward responsible AI. Inf Fus 58:82\u2013115\nAsan O, Bayrak AE, Choudhury A (2020) Artificial intelligence and human trust in healthcare: focus on clinicians. J Med Internet Res 22(6):e15154\nBahrammirzaee A (2010) A comparative survey of artificial intelligence applications in finance: artificial neural networks, expert system, and hybrid intelligent systems. Neural Comput Applic 19:1165\u20131195.https://doi.org/10.1007/s00521-010-0362-z\nBelghachi M (2023) A review of explainable artificial intelligence methods, applications, and challenges. Indones J Electr Eng Inf (IJEEI) 11(4):1007\u20131024\nBench-Capon TJ (2014) Knowledge representation: an approach to artificial intelligence (vol 32). Elsevier, Publisher, pp 220\nBharati S, Mondal MRH, Podder P (2023) A review on explainable artificial intelligence for healthcare: Why, How, and When? IEEE Transact Artif Intell 5(4):1429\u20131442.https://doi.org/10.1109/TAI.2023.3266418\nBhatt U, Xiang A, Sharma S, Weller A, Taly A, Jia Y, Ghosh J, Puri R, Moura JM, Eckersley P (2020) Explainable machine learning in deployment. In: Proceedings of the 2020 conference on fairness, accountability, and transparency (pp 648\u2013657).\nBarria-Pineda J (2020). Exploring the need for transparency in educational recommender systems. In: Proceedings of the 28th ACM conference on user modeling, adaptation and personalization (pp 376\u2013379).\nBibri SE, Krogstie J, Kaboli A, Alahi A (2024) Smarter eco-cities and their leading-edge artificial intelligence of things solutions for environmental sustainability: a comprehensive systematic review. Environ Sci Ecotechnology 19:100330\nBobek S, Kuk M, Szel\u0105\u017cek M, Nalepa GJ (2022) Enhancing cluster analysis with explainable AI and multidimensional cluster prototypes. IEEE Access 10:101556\u2013101574\nBorgonovo E, Plischke E (2016) Sensitivity analysis: a review of recent advances. Eur J Oper Res 248(3):869\u2013887\nBorys K, Schmitt YA, Nauta M, Seifert C, Kr\u00e4mer N, Friedrich CM, Nensa F (2023) Explainable AI in medical imaging: an overview for clinical practitioners\u2013beyond saliency-based XAI approaches. Eur J Radiol 162:110786\nBurzykowski T, Rousseau AJ, Geubbelmans M, Valkenborg D (2023) Introduction to machine learning. Am J Orthod Dentofac Orthop 163(5):732\u2013734\nCartolano A, Cuzzocrea A, Pilato G (2024). Agriculture Environ.https://doi.org/10.1007/s11042-023-17978-z\nChakraborty S, Tomsett R, Raghavendra R, Harborne D, Alzantot M, Cerutti F, Gurram, P (2017) Interpretability of deep learning models: a survey of results. In 2017 IEEE smartworld, ubiquitous intelligence & computing, advanced & trusted computed, scalable computing & communications, cloud & big data computing, Internet of people and smart city innovation (smartworld/SCALCOM/UIC/ATC/CBDcom/IOP/SCI) (pp 1\u20136). IEEE.\nChamola V, Hassija V, Sulthana AR, Ghosh D, Dhingra D, Sikdar B (2023) A review of trustworthy and explainable artificial intelligence (xai). IEEE Access.\nChang-Brahim I, Koppensteiner LJ, Beltrame L, Bodner G, Saranti A, Salzinger J, Molin EM (2024) Reviewing the essential roles of remote phenotyping, GWAS and explainable AI in practical marker-assisted selection for drought-tolerant winter wheat breeding. Front Plant Sci 15:1319938\nChaudhary M, Gaur L, Singh G, Afaq A (2024) Introduction to explainable AI (XAI) in e-commerce. In: Role of explainable artificial intelligence in e-commerce (pp 1\u201315). Cham: Springer Nature Switzerland.\nChoo J, Liu S (2018) Visual analytics for explainable deep learning. IEEE Comput Graphics Appl 38(4):84\u201392\nChowdhary K, Chowdhary KR (2020). Natural language processing. Fundamentals of artificial intelligence, 603\u2013649.\nChromik M, Eiband M, Buchner F, Kr\u00fcger A, Butz A (2021) I think I get your point, AI! The illusion of explanatory depth in explainable AI. In: 26th International Conference on Intelligent User Interfaces (pp. 307\u2013317).\nCohen SN, Snow D, Szpruch L (2021) Black-box model risk in finance. arXiv preprintarXiv:2102.04757, 10.\nCortez P, Embrechts MJ (2013) Using sensitivity analysis and visualization techniques to open black box data mining models. Inf Sci 225:1\u201317\nDazeley R, Vamplew P, Foale C, Young C, Aryal S, Cruz F (2021) Levels of explainable artificial intelligence for human-aligned conversational explanations. Artif Intell 299:103525\nDel Ser J, Barredo-Arrieta A, D\u00edaz-Rodr\u00edguez N, Herrera F, Saranti A, Holzinger A (2024) On generating trustworthy counterfactual explanations. Inf Sci 655:119898\nDeldjoo Y, Nazary F, Ramisa A, Mcauley J, Pellegrini G, Bellogin A, Noia TD (2023) A review of modern fashion recommender systems. ACM Comput Surv 56(4):1\u201337\nD\u00edaz-Rodr\u00edguez N, Del Ser J, Coeckelbergh M, de Prado ML, Herrera-Viedma E, Herrera F (2023) Connecting the dots in trustworthy artificial intelligence: from AI principles, ethics, and key requirements to responsible AI systems and regulation. Inf Fus 99:101896\nDwivedi R, Dave D, Naik H, Singhal S, Omer R, Patel P, Ranjan R (2023) Explainable AI (XAI): core ideas, techniques, and solutions. ACM Comput Surv 55(9):1\u201333\nEberhard K (2023) The effects of visualization on judgment and decision-making: a systematic literature review. Manage Rev Quart 73(1):167\u2013214.https://doi.org/10.1007/s11301-021-00235-8\nEhsan U, Passi S, Liao QV, Chan L, Lee IH, Muller M, Riedl MO (2024) The who in XAI: how AI background shapes perceptions of AI explanations. In: Proceedings of the CHI conference on human factors in computing systems (pp 1\u201332).\nEl Makhloufi A (2023) AI application in transport and logistics: opportunities and challenges (An Exploratory Study).\nEli-chukwu N (2020) Applications of Artificial Intelligence in Agriculture\u202f: A Review. July.https://doi.org/10.48084/etasr.2756\nElton DC (2020) Self-explaining AI as an alternative to interpretable AI. In: artificial general intelligence: 13th international conference, AGI 2020, St. Petersburg, Russia, September 16\u201319, 2020, Proceedings 13 (pp. 95\u2013106). Springer International Publishing\nEnnab M, Mcheick H (2022) Designing an interpretability-based model to explain the artificial intelligence algorithms in healthcare. Diagnostics 12(7):1557\nFanni SC, Febi M, Aghakhanyan G, Neri E (2023) Natural language processing. Introduction to artificial intelligence. Springer International Publishing, Cham, pp 87\u201399\nFelzmann H, Fosch-Villaronga E, Lutz C, Tam\u00f2-Larrieux A (2020) Towards transparency by design for artificial intelligence. Sci Eng Ethics 26(6):3333\u20133361\nFern\u00e1ndez Llorca D, G\u00f3mez E (2021) Trustworthy autonomous vehicles. Publications Office of the European Union, Luxembourg, p 30942\nFresz B, Dubovitskaya E, Brajovic D, Huber MF, Horz C (2024) How should AI decisions be explained? Requirements for explanations from the perspective of European Law. In: Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society (Vol. 7, pp. 438\u2013450).\nFritz-Morgenthal S, Hein B, Papenbrock J (2022) Financial risk management and explainable, trustworthy, responsible AI. Front Artif Intell 5:779799\nFuhrman JD, Gorre N, Hu Q, Li H, El Naqa I, Giger ML (2022) A review of explainable and interpretable AI with applications in COVID-19 imaging. Med Phys 49(1):1\u201314\nGardezi M, Joshi B, Rizzo DM, Ryan M, Prutzer E, Brugler S, Dadkhah A, Ryan M (2023). Artificial intelligence in farming\u202f: Challenges and opportunities for building trust opportunities for improving. March, 1\u201312.https://doi.org/10.1002/agj2.21353\nGe Y, Liu S, Fu Z, Tan J, Li Z, Xu S, Li Y, Xian Y, Zhang Y (2022) A survey on trustworthy recommender systems. arXiv preprintarXiv:2207.12515.\nGelfond M, Kahl Y (2014). Knowledge representation, reasoning, and the design of intelligent agents: The answer-set programming approach. Cambridge University Press.\nGerke S, Minssen T, Cohen G (2020) Ethical and legal challenges of artificial intelligence-driven healthcare. In Artificial intelligence in healthcare (pp 295\u2013336). Academic Press.\nGong J, Yu Q, Li T, Liu H, Zhang J, Fan H, Jin D, Li Y\u00a0(2023). Demo: scalable digital twin system for mobile networks with generative AI. In: Proceedings of the 21st annual international conference on mobile systems, applications and services (MobiSys\u2019 23), Association for computing machinery.https://doi.org/10.1145/3581791.3597297.\nGovers FX (2018) Artificial intelligence for robotics: Build intelligent robots that perform human tasks using AI techniques. Packt Publishing Ltd.\nGrm K, \u0160truc V, Artiges A, Caron M, Ekenel HK (2018) Strengths and weaknesses of deep learning models for face recognition against image degradations. It Biometrics 7(1):81\u201389\nGuidotti R, Monreale A, Ruggieri S, Pedreschi D, Turini F, Giannotti F (2018). Local rule-based explanations of black box decision systems. arXiv preprintarXiv:1805.10820.\nHagras H (2018) Toward human-understandable, explainable AI. Computer 51(9):28\u201336\nHanif A, Zhang X, Wood S (2021) A survey on explainable artificial intelligence techniques and challenges. In: 2021 IEEE 25th international enterprise distributed object computing workshop (EDOCW) (pp. 81\u201389). IEEE.\nHanif PM, Soni M, Soni A (2023) A study on the role of artificial intelligence in e-commerce. J Emerg Technol Innov Res 10(9):105\u2013110\nHashemi M (2023) Who wants what and how: a mapping function for explainable artificial intelligence. arXiv preprintarXiv:2302.03180.\nHeng YS, Subramanian P (2022) A systematic review of machine learning and explainable artificial intelligence (XAI) in credit risk modelling. In: Proceedings of the future technologies conference (pp 596\u2013614). Cham: Springer International Publishing.\nHolstein K, Doroudi S (2022) Equity and artificial intelligence in education. In: The ethics of artificial intelligence in education (pp 151\u2013173) Routledge.\nH\u00fcllmann JA (2022) Explainable AI in farming\u202f: configurations of human-AI joint decision-making. 1\u20134.\nIbrahim R, Shafiq MO (2023) Explainable convolutional neural networks: a taxonomy, review, and future directions. ACM Comput Surv 55(10):1\u201337\nIkegwu AC, Nweke FH, Anikwe CV (2024) Recent trends in computational intelligence for educational big data analysis. Iran J Comput Sci 7(1):103\u2013129.https://doi.org/10.1007/s42044-023-00158-5\nIkegwu AC, Nweke HF, Anikwe CV, Alo UR, Okonkwo OR (2022) Big data analytics for data-driven industry: a review of data sources, tools, challenges, solutions, and research directions. Clust Comput 25(5):3343\u20133387\nIkegwu AC, Nweke HF, Mkpojiogu E, Anikwe CV, Igwe SA, Alo UR (2024) Recently emerging trends in big data analytic methods for modeling and combating climate change effects. Energy Inf 7(1):6\nJan Z, Ahamed F, Mayer W, Patel N, Grossmann G, Stumptner M, Kuusk A (2023) Artificial intelligence for industry 4.0: systematic review of applications, challenges, and opportunities. Expert Syst Appl 216:119456\nJang Y, Choi S, Jung H, Kim H (2022) Practical early prediction of students\u2019 performance using machine learning and eXplainable AI. Educ Inf Technol 27(9):12855\u201312889\nJia J, Liang W, Liang Y (2023) A review of hybrid and ensemble in deep learning for natural language processing. arXiv preprintarXiv:2312.05589.\nKale A, Nguyen T, Harris FC Jr, Li C, Zhang J, Ma X (2023) Provenance documentation to enable explainable and trustworthy AI: a literature review. Data Intell 5(1):139\u2013162\nKaminski ME (2021) The right to explanation, explained. In: Research Handbook on Information Law and Governance (pp 278\u2013299). Edward Elgar Publishing.\nKaushik R (2023) Explainability in machine learning: bridging the gap between model complexity and interpretability. Edu Journal of International Affairs and Research, ISSN:2583\u20139993, 2(4), 57\u201363.\nKesari A, Sele D, Ash E, Bechtold S (2024) A Legal framework for explainable artificial intelligence. Center for Law & Economics Working Paper Series, 9.\nKhosravi H, Shum SB, Chen G, Conati C (2022) Explainable artificial intelligence in education computers and education\u202f: artificial intelligence explainable artificial intelligence in education Dragan Ga \u02c7. May.https://doi.org/10.1016/j.caeai.2022.100074\nKhrais LT (2020) Role of artificial intelligence in shaping consumer demand in E-commerce. Future Internet 12(12):226\nKosasih E, Papadakis E, Baryannis G, Brintrup A (2023). Explainable artificial intelligence in supply chain management: a systematic review of neurosymbolic approaches. Int J Prod Res.\nKrig S (2014) Computer vision metrics: survey, taxonomy, and analysis (p 508). Springer nature.\nKrishnamoorthy CS, Rajeev S (2018) Artificial intelligence and expert systems for engineers. CRC Press\nKulesza T, Stumpf S, Herlocker J (2015) Designing explanatory interfaces. Proceedings of the 2015 international conference on intelligent user interfaces. Link\nKumar Y, Jain Y (2012) Research aspects of an expert system. Int J Comput Bus Res 1(11).https://api.semanticscholar.org/CorpusID:61842788\nKu\u017aniacki B, Almada M, Tyli\u0144ski K, G\u00f3rski \u0141, Winogradska B, Zeldenrust R (2022) Towards explainable artificial intelligence (XAI) in tax law: the need for a minimum legal standard. World Tax J 14:573\u2013616\nKuznietsov A, Gyevnar B, Wang C, Peters S, Albrecht SV (2024). Explainable AI for safe and trustworthy autonomous driving: a systematic review. arXiv preprintarXiv:2402.10086.\nLakshmi V, Corbett J (2023). Using AI to improve sustainable agricultural practices: a literature review and research agenda.\nLawrence B (2019) The role of legal scholars in the development of artificial intelligence. Harv Law Rev Forum 132:157\u2013167\nLee A (2024) Knowledge representation and reasoning in AI: analyzing different approaches to knowledge representation and reasoning in artificial intelligence systems. J Artif Intell Res 4(1):14\u201329\nLee AVY, Koh E, Looi CK (2023) AI in education and learning analytics in Singapore: an overview of key projects and initiatives. Inf Technol Edu Learn 3(1):Inv-p001\nLim S (2021) Judicial decision-making and explainable artificial intelligence: a reckoning from first principles. Singap Acad Law J 33:280\u2013314\nLinheiro ES, Shinde GR, Mahalle PN, Mirajkar R (2023) Explainable AI (XAI) for agriculture. Industry 4.0 convergence with AI, IoT, big data and cloud computing: fundamentals, challenges and applications, 161.\nLisboa PJ, Saralajew S, Vellido A, Fern\u00e1ndez-Domenech R, Villmann T (2023) The coming of age of interpretable and explainable machine learning models. Neurocomputing 535:25\u201339\nLi M, Wan Z, Zou T, Shen Z, Li M, Wang C, Xiao X (2024) Artificial intelligence enabled self-powered wireless sensing for smart industry. Chem Eng J 492:152417\nLiu J, Chen K, Lyu W (2024) Embracing artificial intelligence in the labour market: the case of statistics. Humanit Soc Sci Commun 11(1):1.https://doi.org/10.1057/s41599-024-03557-6\nLi T, Li Y (2023) Artificial intelligence for reducing the carbon emissions of 5G networks in China. Nat Sustain 6(12):1522\u20131523\nLiu Q, Pinto JD, Paquette L (2024) Applications of explainable ai (xai) in education. In Trust and Inclusion in AI-Mediated Education: Where Human Learning Meets Learning Machines (pp. 93\u2013109). Cham: Springer Nature Switzerland.\nLong S, Huang W, Wang J, Liu J, Gu Y, Wang Z (2024) A fixed-time consensus control with prescribed performance for multi-agent systems under full-state constraints. IEEE Transact Autom Sci Eng\nLove PE, Fang W, Matthews J, Porter S, Luo H, Ding L (2023) Explainable artificial intelligence (XAI): precepts, models, and opportunities for research in construction. Adv Eng Inform 57:102024\nLythe M, de Cos GM, Mingallon M, Lensen A, Galloway C, Knox D, Auvaa S, Kumarasinghe K (2023) Explainable AI\u2013building trust through understanding.\nMahbooba B, Timilsina M, Sahal R, Serrano M (2021) Explainable artificial intelligence (XAI) to enhance trust management in intrusion detection systems using decision tree model. Complexity 2021:1\u201311\nMajumdar S (2023) Fairness, explainability, privacy, and robustness for trustworthy algorithmic decision-making. In Big data analytics in chemoinformatics and bioinformatics (pp 61\u201395). Elsevier.\nMana AA, Allouhi A, Hamrani A, Rahman S, el Jamaoui I, Jayachandran K (2024) Sustainable AI-based production agriculture: exploring AI applications and implications in agricultural practices. Smart Agric Technol 17:100416\nMatcov A (2024) Explainable AI in Credit Risk Assessment for External Customers (Bachelor\u2019s thesis, University of Twente).\nMehta S, Rogers A, Gilbert TK (2023) Dynamic Documentation for AI Systems. arXiv preprintarXiv:2303.10854.\nMinh D, Wang HX, Li YF, Nguyen TN (2022) Explainable artificial intelligence: a comprehensive review. Artificial Intelligence Review, 1\u201366.\nMittelstadt B, Russell C, Wachter S (2019) Explaining explanations in AI. In Proceedings of the conference on fairness, accountability, and transparency (pp 279\u2013288).\nMohammed AA, Ambak K, Mosa AM, Syamsunur D (2019) Expert system in engineering transportation: a review. J Eng Sci Technol 14(1):229\u2013252\nMontavon G, Samek W, M\u00fcller KR (2018) Methods for interpreting and understanding deep neural networks. Digital Signal Process 73:1\u201315\nNagahisarchoghaei M, Nur N, Cummins L, Nur N, Karimi MM, Nandanwar S, Rahimi S (2023) An empirical survey on explainable ai technologies: recent trends, use-cases, and categories from technical and application perspectives. Electronics 12(5):1092\nNallakaruppan MK, Balusamy B, Shri ML, Malathi V, Bhattacharyya S (2024) An Explainable AI framework for credit evaluation and analysis. Appl Soft Comput 153:111307\nNixon M, Aguado A (2019) Feature extraction and image processing for computer vision. Academic press.\nNoguerol TM, Paulano-Godino F, Mart\u00edn-Valdivia MT, Menias CO, Luna A (2019) Strengths, weaknesses, opportunities, and threats analysis of artificial intelligence and machine learning applications in radiology. J Am Coll Radiol 16(9):1239\u20131247\nNoorbakhsh-Sabet N, Zand R, Zhang Y, Abedi V (2019) Artificial intelligence transforms the future of health care. Am J Med 132(7):795\u2013801\nNott G (2017) Explainable artificial intelligence: Cracking open the black box of AI. Computer world, 4.\nNwakanma CI, Ahakonye LA, Njoku JN, Odirichukwu JC, Okolie SA, Uzondu C, Ndubuisi Nweke CC, Kim D (2023) Applied sciences explainable artificial intelligence ( XAI ) for intrusion detection and mitigation in intelligent connected vehicles a review. Appl Sci 13:1252\nNwosu AC, Sturgeon B, McGlinchey T, Goodwin CD, Behera A, Mason S, Payne TR (2019) Robotic technology for palliative and supportive care: strengths, weaknesses, opportunities and threats. Palliat Med 33(8):1106\u20131113\nO\u2019Mahony N, Campbell S, Carvalho A, Harapanahalli S, Hernandez GV, Krpalkova L, Riordan D, Walsh J (2020) Deep learning vs. traditional computer vision. In Advances in Computer Vision: Proceedings of the 2019 Computer Vision Conference (CVC), Volume 1 1 (pp. 128\u2013144). Springer International Publishing.\nPanayides AS, Amini A, Filipovic ND, Sharma A, Tsaftaris SA, Young A, Pattichis CS (2020) AI in medical imaging informatics: current challenges and future directions. IEEE J Biomed Health Info 24(7):1837\u20131857\nPawar U, O\u2019Shea D, Rea S, O\u2019Reilly R\u00a0(2020). Incorporating Explainable Artificial Intelligence (XAI) to aid the Understanding of Machine Learning in the Healthcare Domain. InAics(pp 169\u2013180).\nPlass M, Kargl M, Kiehl TR, Regitnig P, Gei\u00dfler C, Evans T, M\u00fcller H (2023) Explainability and causability in digital pathology. J Pathol: Clin Res 9(4):251\u2013260\nPoch\u00e9 A, Hervier L, Bakkay MC (2023). Natural Example-Based Explainability: a Survey. In World Conference on explainable Artificial Intelligence (pp 24\u201347). Cham: Springer Nature Switzerland.\nPosadas BB, Ogunyiola A, Niewolny K (2023) Socially responsible AI assurance in precision agriculture for farmers and policymakers. In AI Assurance (pp 473\u2013499). Academic Press.\nPoursabzi-Sangdeh F, Goldstein DG, Hofman JM, Wortman Vaughan JW, Wallach H (2021) Manipulating and measuring model interpretability. In: Proceedings of the 2021 CHI conference on human factors in computing systems (pp. 1\u201352).\nPushkarna M, Zaldivar A, Kjartansson O (2022) Data cards: purposeful and transparent dataset documentation for responsible AI. In: Proceedings of the 2022 ACM conference on fairness, accountability, and transparency (pp 1776\u20131826).\nRachha A, Seyam M (2023) Explainable AI in education\u202f: current trends, challenges and opportunities. SoutheastCon 2023.https://doi.org/10.1109/SoutheastCon51012.2023.10115140\nRagno A, La Rosa B, Capobianco R (2022) Prototype-based interpretable graph neural networks. IEEE Transactions on Artificial Intelligence.\nRane N, Choudhary S, Rane J (2023) Explainable artificial intelligence (XAI) approaches for transparency and accountability in financial decision-making. Available at SSRN 4640316.\nRasheed MH, Khalid J, Ali A, Rasheed MS, Ali K (2024) Human resource analytics in the era of artificial intelligence: leveraging knowledge towards organizational success in Pakistan. J Chin Hum Resour Manag 15(3):3\u201320.https://doi.org/10.47297/wspchrmWSP2040-800501.20241503\nRong Y, Xu Z, Liu J, Liu H, Ding J, Liu X, Gao J (2022) Du-bus: a realtime bus waiting time estimation system based on multi-source data. IEEE Transact Intell Transp Syst 23(12):24524\u201324539\nRajan K, Saffiotti A (2017) Towards a science of integrated AI and robotics. Artif Intell 247:1\u20139\nReshamwala A, Mishra D, Pawar P (2013) Review on natural language processing. IRACST Eng Sci Technol: Int J (ESTIJ) 3(1):113\u2013116\nRetzlaff CO, Angerschmid A, Saranti A, Schneeberger D, Roettger R, Mueller H, Holzinger A (2024) Post-hoc vs ante-hoc explanations: xAI design guidelines for data scientists. Cogn Syst Res 86:101243\nRyo M (2022) Explainable artificial intelligence and interpretable machine learning for agricultural data analysis. Artif Intell Agric 6:257\u2013265\nSabrina F, Sohail S, Farid F, Jahan S, Ahamed F, Gordon S (2022) An interpretable artificial intelligence-based smart agriculture system. Comput Mater Continua 72:3777\u20133797\nSadeghi K, Ojha D, Kaur P, Mahto RV, Dhir A (2024) Explainable artificial intelligence and agile decision-making in supply chain cyber resilience. Decis Support Syst 180:114194\nSaeed W, Omlin C (2023) Explainable AI (XAI): a systematic meta-survey of current challenges and future opportunities. Knowl-Based Syst 263:110273\nSamek W, Wiegand T, M\u00fcller KR (2017) Explainable artificial intelligence: understanding, visualizing, and interpreting deep learning models. arXiv preprintarXiv:1708.08296.\nSarker IH (2022) AI-based modeling: techniques, applications and research issues towards automation, intelligent and smart systems. SN Computer Science 3(2):158\nSchoenherr JR, Abbas R, Michael K, Rivas P, Anderson TD (2023) Designing AI using a human-centered approach: Explainability and accuracy toward trustworthiness. IEEE Transact Technol Soc 4(1):9\u201323\nSchoonderwoerd TA, Jorritsma W, Neerincx MA, Van Den Bosch K (2021) Human-centered XAI: developing design patterns for explanations of clinical decision support systems. Int Lodg Ind: Orig Last Front 154:279\u2013313\nStepin I, Alonso JM, Catala A, Pereira-Fari\u00f1a M (2021) A survey of contrastive and counterfactual explanation generation methods for explainable artificial intelligence. IEEE Access 9:11974\u201312001\nStiglic G, Kocbek P, Fijacko N, Zitnik M, Verbert K, Cilar L (2020) Interpretability of machine learning-based prediction models in healthcare. Wiley Interdiscip Rev: Data Min Knowl Discov 10(5):e1379\nSunitha B, Kiran BK (2023) A systematic review of recommendation systems: applications and challenges. Int J Intell Syst Appl Eng 11(2):826\u2013846\nSutherland A, Wright D (2020) The Governance of Artificial Intelligence in the Legal Sector. AI & Society, 35(1), 119\u2013133. Link\nTaghikhah F, Voinov A, Filatova T, Polhill JG (2022) Machine-assisted agent-based modeling: opening the black box. J Comput Sci 64:101854\nTalaat FM, Aljadani A, Badawy M, Elhosseini M (2024) Toward interpretable credit scoring: integrating explainable artificial intelligence with deep learning for credit card default prediction. Neural Comput Appl 36(9):4847\u20134865\nTorfi A, Shirvani RA, Keneshloo Y, Tavaf N, Fox EA (2020) Natural language processing advancements by deep learning: A survey. arXiv preprintarXiv:2003.01200.\nTucci V, Saary J, Doyle TE\u00a0(2022) Factors influencing trust in medical artificial intelligence for healthcare professionals: A narrative review. J Med Artif Intell, 5.\nValavanis KP, Saridis GN (2012) Intelligent robotic systems: theory, design, and applications (vol 182). Springer Science & Business Media.\nVale D, El-Sharif A, Ali M (2022) Explainable artificial intelligence (XAI) post-hoc explainability methods: risks and limitations in non-discrimination law. AI and Ethics 2(4):815\u2013826\nVilone G, Rizzo L, Longo L (2020) A comparative analysis of rule-based, model-agnostic methods for explainable artificial intelligence.\nVinod B (2022) Artificial intelligence and emerging technologies in hospitality. revenue management in the lodging industry: origins to the last frontier, 279\u2013313.\nVollert S, Atzmueller M, Theissler A (2021) Interpretable Machine Learning: A brief survey from the predictive maintenance perspective. In 2021 26th IEEE international conference on emerging technologies and factory automation (ETFA) (pp 01\u201308). IEEE.\nVoj\u00ed\u0159 S, Kliegr T (2020) Editable machine learning models? A rule-based framework for user studies of explainability. Adv Data Anal Classif 14(4):785\u2013799\nVoulodimos A, Doulamis N, Doulamis A, Protopapadakis E (2018). Deep learning for computer vision: A brief review. Computational intelligence and neuroscience, 2018.\nVyas B (2023) Explainable AI: assessing methods to make AI systems more transparent and interpretable. Int J New Med Stud: Int Peer Rev Sch Index J 10(1):236\u2013242\nWalker CM, Agarwal V, Lin L, Hall AC, Hill RA, Laurids R, Mortenson TJ, Lybeck NJ (2023) Explainable Artificial Intelligence Technology for Predictive Maintenance (No. INL/RPT-23-74159-Rev000). Idaho National Laboratory (INL), Idaho Falls, ID (United States).\nWang E, Yang Y, Wu J, Liu W, Wang X (2018) An efficient prediction-based user recruitment for mobile crowdsensing. EEE Transact Mob Comput 17(1):16\u201328\nWang P, Wei Z, Qi H, Wan S, Xiao Y, Sun G, Zhang Q (2023) Mitigating poor data quality impact with federated unlearning for human-centric metaverse. IEEE J Sel Areas Commun\nWang XS, Ryoo JHJ, Bendle N, Kopalle PK (2021) The role of machine learning analytics and metrics in retailing research. J Retail 97(4):658\u2013675\nWang Y, Chandrasekaran J, Haberkorn F, Dong Y, Gopinath M, Batarseh FA (2022) Deepfarm: AI-driven management of farm production using explainable causality. In 2022 IEEE 29th annual software technology conference (STC) (pp 27\u201336). IEEE.\nWeber L, Lapuschkin S, Binder A, Samek W (2023) Beyond explaining: Opportunities and challenges of XAI-based model improvement. Inf Fus 92:154\u2013176\nWeber P, Carl KV, Hinz O (2023) Applications of explainable artificial intelligence in finance\u2014a systematic review of finance, information systems, and computer science literature. Management Review Quarterly, 1\u201341.\nWu L, Long Y, Gao C, Wang Z, Zhang Y (2023) MFIR: multimodal fusion and inconsistency reasoning for explainable fake news detection. Inf Fus 100:101944\nZhang S, Li T, Hui S, Li G, Liang Y, Yu L, Jin D, Li Y (2023) Deep Transfer Learning for City-scale Cellular Traffic Generation through Urban Knowledge Graph. Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 4842\u20134851.https://doi.org/10.1145/3580305.3599801\nWu K, Chi K (2023) Enhanced e-commerce customer engagement: A comprehensive three-tiered recommendation system. Journal of Knowledge Learning and Science Technology ISSN: 2959\u20136386 (online), 2(3), 348\u2013359.\nYang Y, Zhuang Y, Pan Y (2021) Multiple knowledge representation for big data artificial intelligence: framework, applications, and case studies. Front Inf Technol Electron Eng 22(12):1551\u20131558\nZacharias J, von Zahn M, Chen J, Hinz O (2022) Designing a feature selection method based on explainable artificial intelligence. Electron Mark 32(4):2159\u20132184\nZhang Y, Weng Y, Lund J (2022) Applications of explainable artificial intelligence in diagnosis and surgery. Diagnostics 12(2):237\nZhu C (2024) Computational intelligence-based classification system for the diagnosis of memory impairment in psychoactive substance users. J Cloud Comput.https://doi.org/10.1186/s13677-024-00675-z\nZou X, Yuan J, Shilane P, Xia W, Zhang H, Wang X (2022) From hyper-dimensional structures to linear structures: maintaining deduplicated data\u2019s locality. ACM Transact Storage.https://doi.org/10.1145/3507921\nZhuang YT, Wu F, Chen C, Pan YH (2017) Challenges and opportunities: from big data to knowledge in AI 2.0. Front Inf Technol Electron Eng 18:3\u201314\nFunding\nAuthor information\nAuthors and Affiliations\nCentre for Atmospheric Research, National Space Research and Development Agency, Anyigba, Nigeria\nComputer Engineering Department, Faculty of Engineering, Veritas University, Abuja, Nigeria\nSoftware Engineering Department, Faculty of Natural and Applied Sciences, Veritas University, Abuja, Nigeria\nComputer Science Department, Faculty of Physical Sciences, University of Nigeria, Nsukka (UNN), Nigeria\nDaniel Enemona Mathew,\u00a0Deborah Uzoamaka Ebem,\u00a0Pamela Eberechukwu Ukeoma\u00a0&\u00a0Ngozi Fidelia Dibiaezue\nContributions\nDEM coined the title, surfed the internet for materials, and participated in manuscript preparation especially Sect. 2, 3 and 4. ACI drafted out the structures, wrote \u201cThe introduction, and conclusion sections, and participated in the reviewing of the entire manuscript sections. DUE participated in the drafting, grammatical check, correction, and editing. PEU screened the materials and participated in writing the \u201cExplanatory Depth of Explainable AI\u201d section. In addition, NFD participated in the manuscript draft, proofreading, and reviewing of the entire manuscript.\nCorresponding authors\nEthics declarations\nConflict of interest\nEthical Approval and Consent to Participate\nConsent for Publication\nAdditional information\nPublisher's Note\nSpringer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.\nRights and permissions\nOpen AccessThis article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article\u2019s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article\u2019s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visithttp://creativecommons.org/licenses/by-nc-nd/4.0/.\nAbout this article\nCite this article\nMathew, D.E., Ebem, D.U., Ikegwu, A.C.et al.Recent Emerging Techniques in Explainable Artificial Intelligence to Enhance the Interpretable and Understanding of AI Models for Human.Neural Process Lett57, 16 (2025). https://doi.org/10.1007/s11063-025-11732-2\nShare this article\nKeywords\nSearch\nNavigation\nDiscover content\nPublish with us\nProducts and services\nOur brands", "cleaned_text_custom_parser_2": "\n\n \n \n Skip to main content\n \n

\n Advertisement\n

\n Recent Emerging Techniques in Explainable Artificial Intelligence to Enhance the Interpretable and Understanding of AI Models for Human\n

\n \n Open access\n \n
\n Published:\n \n 07 February 2025\n \n

\n Volume\u00a057\n , article\u00a0number\n 16\n , (\n 2025\n )\n
\n \n Cite this article\n \n

\n \n Download PDF\n \n

\n You have full access to this\n \n open access\n \n article\n

\n \n Neural Processing Letters\n \n \n Aims and scope\n \n \n Submit manuscript\n \n

\n Recent Emerging Techniques in Explainable Artificial Intelligence to Enhance the Interpretable and Understanding of AI Models for Human\n \n Download PDF\n \n

\n Abstract\n

\n Recent advancements in Explainable Artificial Intelligence (XAI) aim to bridge the gap between complex artificial intelligence (AI) models and human understanding, fostering trust and usability in AI systems. However, challenges persist in comprehensively interpreting these models, hindering their widespread adoption. This study addresses these challenges by exploring recently emerging techniques in XAI. The primary problem addressed is the lack of transparency and interpretability in AI models to humanity for institution-wide use, which undermines user trust and inhibits their integration into critical decision-making processes. Through an in-depth review, this study identifies the objectives of enhancing the interpretability of AI models and improving human understanding of their decision-making processes. Various methodological approaches, including post-hoc explanations, model transparency methods, and interactive visualization techniques, are investigated to elucidate AI model behaviours. We further present techniques and methods to make AI models more interpretable and understandable to humans including their strengths and weaknesses to demonstrate promising advancements in model interpretability, facilitating better comprehension of complex AI systems by humans. In addition, we provide the application of XAI in local use cases. Challenges, solutions, and open research directions were highlighted to clarify these compelling XAI utilization challenges. The implications of this research are profound, as enhanced interpretability fosters trust in AI systems across diverse applications, from healthcare to finance. By empowering users to understand and scrutinize AI decisions, these techniques pave the way for more responsible and accountable AI deployment.\n

\n Similar content being viewed by others\n

\n \n Introduction to Explainable AI\n \n

\n Chapter\n \u00a9 2024\n

\n \n Study on\u00a0the\u00a0Helpfulness of\u00a0Explainable Artificial Intelligence\n \n

\n Chapter\n \u00a9 2024\n

\n \n Transformative impact of explainable artificial intelligence: bridging complexity and trust\n \n

\n Article\n Open access\n 09 May 2025\n

\n Explore related subjects\n

\n Discover the latest articles and news from researchers in related subjects, suggested using machine learning.\n

\n \n Computational Intelligence\n \n
\n \n Intelligence Augmentation\n \n
\n \n Machine Learning\n \n
\n \n Philosophy of Artificial Intelligence\n \n
\n \n Symbolic AI\n \n
\n \n Artificial Intelligence\n \n

\n \n Use our pre-submission checklist\n \n

\n Avoid common mistakes on your manuscript.\n

\n 1\n Introduction\n

\n Artificial Intelligence (AI) has played an increasingly significant role in using data collected from diverse sources such as social media, sensors, education, multimedia, and text data, to make necessary decisions that have helped humans [\n \n 67\n \n ,\n \n 68\n \n ,\n \n 69\n \n ]. Explainable Artificial Intelligence (XAI) stems from the increasing integration of AI models in a way that allows humans to understand, interpret, and trust the decisions and outputs generated by AI systems because these systems have become more sophisticated and pervasive. There is a growing need for transparency and interpretability in decision-making processes. With confidence in the deployment of AI by industries and governments, users are seeking enhanced transparency facilitated by XAI systems where the author introduces XAI concepts and provides an overview of areas requiring additional investigation, such as type-2 fuzzy logic systems [\n \n 58\n \n ]. The aim guarantees these systems are comprehensible and analyzable by users without specialized expertise. Understanding how AI models arrive at specific conclusions or recommendations is important for building trust and acceptance among users, stakeholders, and society. Ensuring that AI systems are accurate and interpretable helps address concerns related to bias, fairness, and unintended consequences [\n \n 98\n \n ]. XAI techniques enable practitioners and regulators to assess the ethical implications of AI decisions, identify potential biases, and mitigate any adverse effects which are particularly relevant in fields such as healthcare, finance, manufacturing, and criminal justice, where the consequences of AI decisions can have significant real-world impact [\n \n 123\n \n ].\n

\n Explainable Artificial Intelligence (XAI) lies in its potential to bridge the gap between the complexity of advanced AI models and the need for human understanding and trust. As AI systems continue to play an increasingly pivotal role in critical decision-making processes, it is imperative to enhance their interpretability, making their inner workings more transparent and accessible to users. This not only promotes accountability and ethical use of AI but also addresses concerns related to unintended concerns in empowering individuals, regulators, and stakeholders to comprehend and scrutinize AI models, fostering a responsible and trustworthy integration of artificial intelligence into various aspects of society. In an earlier study by Nott [\n \n 114\n \n ], XAI makes AI more transparent and explainable, addressing the \u201cblack box\u201d problem. However, the results achieved do not clearly explain how or why they arrived at those conclusions. Taghikhah et al. [\n \n 145\n \n ] and Adadi and Berrada [\n \n 2\n \n ] presented \"Cracking open the black box of AI\" refers to the concept of making AI systems more transparent and understandable. AI often operates as a \"black box\" because it can be challenging to interpret how these systems arrive at their decisions or predictions, especially in complex deep learning models and other inner workings of AI systems. Cracking opens the black box of AI, which is essential for building trust in AI systems, ensuring that AI operates ethically, and allowing users to understand and verify AI decisions. Nonetheless, these require much critical attention and development as AI continues to be integrated into various aspects of society [\n \n 155\n \n ].\n

\n To make AI processes more understandable, several key strategies in XAI techniques like model-agnostic interpretability methods can be employed and incorporated to enable users to interpret and comprehend the decisions made by complex AI models [\n \n 37\n \n ]. Also, simpler and more interpretable models whenever possible are used instead of highly complex black-box models for transparency to provide clear and concise explanations for AI outputs, including the factors influencing decisions, and enhance user understanding. Furthermore, fostering transparency in the data used to train these models, addressing biases, and regularly auditing and validating AI systems contribute to the overall interpretability and trustworthiness of AI processes because collaborative efforts between AI developers, domain experts, and end-users in the design and deployment stages also play a crucial role in making AI processes more understandable and aligned with human values and expectations [\n \n 37\n \n ]. Application of XAI in diverse areas such as healthcare, finance, legal, autonomous vehicles, robotics, etc. enhances the interpretability and understanding of AI models for humans. For instance, Amann et al. [\n \n 5\n \n ] utilized XAI to assist clinicians to comprehend and trust complex medical diagnoses and treatment recommendations made by AI systems, ensuring that decisions align with medical expertise. Rane et al. [\n \n 127\n \n ] implored XAI in finance to help in explaining algorithmic trading strategies and risk assessment, enabling financial professionals and regulators to make informed decisions and maintain market transparency. In legal and criminal justice systems, XAI assists in explaining the rationale behind AI-generated predictions, ensuring fairness and accountability [\n \n 90\n \n ]. More so, Fern\u00e1ndez Llorca, and G\u00f3mez [\n \n 46\n \n ] utilized XAI in autonomous vehicles and robotics, to enhance the transparency of decision-making processes, fostering user trust and safety. These applications underscore the importance of XAI in promoting transparency, accountability, and user comprehension across various domains where AI plays a pivotal role.\n

\n Despite the promising potential, Saeed and Omlin [\n \n 136\n \n ] reported that XAI faces several challenges in enhancing the interpretability and understanding of AI models for humans, such as the inherent trade-off between model complexity and interpretability, as highly sophisticated models often sacrifice transparency for performance. Also, developing commonly accepted standards for interpretability poses a significant challenge, as different domains may require tailored explanations, and resolving the issue of context-aware explanations is another obstacle, as the interpretability of AI models should account for the specific needs and knowledge levels of diverse user groups. Striking a balance between accuracy and simplicity in explanation generation remains a persistent challenge, especially in complex, high-dimensional datasets, Kaushik [\n \n 75\n \n ]. Furthermore, Chamola, et al. [\n \n 25\n \n ] reviewed that ensuring that XAI methods effectively handle real-world scenarios, including unforeseen situations and adversarial attacks, is essential for their practical applicability, and overcoming these challenges is imperative for the widespread adoption of XAI, ensuring that AI systems are not only accurate but also transparent, trustworthy, and accessible to a broad range of users.\n

\n Progressively, various studies have attempted to address these peculiar challenges. For instance, Lythe et al. [\n \n 99\n \n ] pinpointed that interpretability goes beyond explainability, aiming to make AI models more accessible to individuals, particularly domain experts unfamiliar with intricate AI techniques, by demystifying complex model architectures. Felzmann et al. [\n \n 45\n \n ] discussed that transparency in AI involves disclosing information about data, algorithms, and processes, enhancing visibility throughout the AI development and deployment stages. Additionally, ongoing research focuses on developing various explainable AI techniques, such as feature attribution methods, rule-based models, and attention mechanisms that unveil influential aspects of input data in model decisions. However, to our knowledge, none of these studies provides a comprehensive approach in solving the inherent issues beclouding the deployment of AI. Ehsan et al. [\n \n 39\n \n ] qualitatively explain how an individual's AI background can shape their interpretations, highlighting differences through the perspectives of appropriation and cognitive heuristics. The study finds that both groups exhibited undue trust in numerical data for different reasons and appreciated various explanations beyond their intended design. These findings, which have significant implications for XAI, reveal that AI-generated explanations can inadvertently lead to negative consequences and potentially harmful manipulation of trust. The authors propose design interventions to address these issues.\n

\n Conversely, this paper addresses a significant research gap in the field of artificial intelligence (AI) because it primarily revolves around the pressing need for AI models to be more transparent and interpretable to humans. Therefore, the contributions of this paper to the current body of knowledge are as follows:\n

\n
\n We provided a comprehensive understanding of the recent emerging techniques in XAI that address an in-depth dimension of explanatory;\n
\n
\n
\n We investigated and evaluated the efficacy of specific techniques in XAI, such as sensitivity analysis, rule-based models, documentation and transparency methods, prototype-based explanations, and visualization techniques;\n
\n
\n
\n Provides the strengths, weaknesses, and features of AI components, including machine learning, natural language processing, computer vision, knowledge representation and reasoning, robotics, and expert systems;\n
\n
\n
\n We explore an in-depth utilization of XAI techniques in diverse local use cases spanning healthcare diagnostics, financial risk assessment, retail and e-commerce, agriculture, education, transport, and logistics;\n
\n
\n
\n We further identified and analyzed the challenges associated with deploying XAI techniques in local use cases across different domains;\n
\n

\n The rest of the paper is organized into the following sections: Sect.\n \n 2\n \n delves into the explanatory depth of XAI. Section\n \n 2.1\n \n outlines techniques and methods to enhance the interpretability and understandability of AI models for humans, while Sect.\n \n 2.2\n \n discusses the understandability of AI components in detail. Furthermore, Sect.\n \n 3\n \n gave an in-depth discussion on the deployment of AI explainable in local use cases while Sect.\n \n 4\n \n covered challenges and solutions of AI explainable in local use cases. Finally, Sect.\n \n 5\n \n handled areas on open research direction.\n

\n 2\n Explanatory Depth of Explainable AI\n

\n In the context of XAI, explanatory depth refers to the extent of detail and clarity in the explanations offered by AI systems. Some works have provided explanatory depth on XAI; Chromik et al. [\n \n 30\n \n ] stated that explanatory depth shows how deeply an AI system can explain its decisions, going beyond seeming or shallow explanations and the concept of explanatory depth is vital because not all AI explanations are equally informative or useful. Some AI systems may provide high-level, imprecise explanations that do not help users understand the underlying reasoning, while others can offer detailed, insightful explanations that enhance user\u2019s trust and comprehension [\n \n 61\n \n ]. So, Hulsen [\n \n 65\n \n ] in his work stated that XAI gives the richness and detail of the explanations provided by AI systems in making AI more understandable and trustworthy, and achieving it involves various techniques and considerations to ensure that the explanations are meaningful and helpful to users. There are various methods used to achieve explanatory depth in XAI and these techniques include, Antoniadi et al. [\n \n 8\n \n ] reviewed that feature importance that is the most influential feature in the decision-making process is pointed out, Eberhard, [\n \n 38\n \n ] extracted that visualization using visual aids shows the decision process and feature interactions, and Mittelstadt et al. [\n \n 106\n \n ] explained that counterfactual explanations by showing what changes in input data would have resulted in a different decision and local verse global explanations that offer insight into how a specific decision was made (local) as well as how the overall model behaves (global). A recent study by Del Ser et al. [\n \n 34\n \n ], introduced a framework for generating counterfactual explanations, formulating the task as a multi-objective optimization problem that balances three goals: plausibility, change intensity, and adversarial effectiveness which uses generative adversarial networks (GANs) to model input distribution, alongside a solver for multi-objective counterfactual discovery, the framework addresses the envisaged peculiar problems. The study demonstrates across six classification tasks using image and 3D data confirm evidence of trade-offs between objectives and consistency of the generated explanations with human understanding, and the framework's ability to reveal concept-based biases and misrepresented attributes within the model's input domain. Therefore, XAI techniques, including feature importance, visualization, and counterfactual explanations, enhance explanatory depth by enabling users to comprehend and trust the decisions made by AI systems.\n

\n Post-hoc explanation is an external method used after a model is trained to clarify its predictions. For instance, Retzlaff et al. [\n \n 132\n \n ] introduced a decision tree that helps navigate key trade-offs in various XAI methods, guiding users in selecting from six popular XAI tools. Post-hoc methods are beneficial for explaining complex models that are not naturally interpretable, such as deep neural networks or ensemble models. Techniques commonly used for post-hoc explanation include feature importance scores, SHapley Additive Explanations (SHAP), and (Local Interpretable Model-agnostic Explanations (LIME), enabling complex models to offer valuable insights into their decision-making processes. Through expert interviews, post-hoc XAI methods will be evaluated for their applicability in real-world contexts. The goal is to enable data scientists and policymakers to choose XAI methods that not only clarify the decision-making process but also enhance user comprehension and interpretation, ultimately fostering the practical adoption of XAI. Vale et al. [\n \n 150\n \n ] alluded to the insights from\u00a0post-hoc explainability methods used to help regulate black-box machine learning. Some of the post-hoc explanations include in healthcare, SHAP helps identify how patient features (such as age, symptoms, or lab results) influence a diagnosis or treatment recommendation, building trust in AI-driven decisions. LIME, on the other hand, explains individual predictions, clarifying why specific diagnoses or risk scores were assigned, which is especially useful for case-specific interpretations. In finance, SHAP clarifies how factors like credit history, income, and debt impact loan approvals or risk assessments, supporting regulatory compliance and customer trust. LIME offers case-specific explanations for individual loan or credit decisions, helping both analysts and customers understand specific outcomes. In agriculture, SHAP explains predictions on crop yield, disease risk, and soil quality by highlighting the impact of factors like weather, soil pH, and fertilizer use, supporting data-driven crop management while LIME offers specific explanations for cases such as high disease risk scores, guiding targeted interventions. In education, SHAP clarifies how factors like attendance, grades, and engagement affect predictions of student performance or dropout risk, helping educators target areas for intervention. LIME provides case-specific insights, allowing educators to understand why particular students are identified as high-risk and address their unique needs. In retail, SHAP highlights key customer features, like purchase history and demographics, that influence behavior predictions, aiding in personalization strategies. LIME provides specific explanations for individual recommendations, allowing retailers to refine algorithms and enhance customer experience. In transport, SHAP identifies factors affecting delivery times, maintenance needs, or fuel efficiency\u2014such as traffic, vehicle age, or route length\u2014helping to boost efficiency. LIME offers explanations for specific delays or routing choices, enabling managers to understand and adjust schedules as needed.\n

\n Zacharias et al. [\n \n 169\n \n ] explained that the feature importance technique seeks to improve explanatory depth by pinpointing the most significant features influencing an AI model's decisions. By determining which features most affect model predictions, users gain insights into the factors driving the AI system's behaviour. It allows users to prioritize their attention and understand the underlying logic of the model so that users can focus on relevant factors and comprehend why certain decisions are made, thereby increasing the transparency and interpretability of the AI system. Visualization techniques play a crucial role in achieving explanatory depth by providing intuitive representations of AI model behaviour by transforming complex model outputs into visually interpretable formats, such as charts, graphs, or heatmaps, users can better understand the patterns and relationships captured by the AI system Minh et al. [\n \n 105\n \n ]. Visualization techniques enable users to explore the inner workings of the model, identify trends, and detect anomalies more effectively. Furthermore, interactive visualizations allow users to manipulate data inputs and observe real-time changes in model predictions, facilitating a deeper understanding of how the AI system responds to different scenarios. In the work of Stepin et al. [\n \n 141\n \n ] counterfactual explanations offer another avenue for achieving explanatory depth in XAI by providing insights into \u201cwhat-if\u201d scenarios by generating hypothetical instances where model predictions change, users can gain a better understanding of the decision boundaries and causal relationships within the AI system which help users explore alternative outcomes and understand how changes in input variables influence model predictions. The goals of Explainable AI are presented in Fig.\n \n 1\n \n .\n

\n \n Fig.\u00a01\n \n

\n Goals of explainable AI\n

\n \n Full size image\n \n \n

\n Figure\n \n 1\n \n illustrates that the primary objective of Explainable Artificial Intelligence (XAI) is to improve the comprehension and acceptance of AI systems. This is achieved by integrating essential principles such as transparency, reliability, causality, usability, privacy, trust, and fairness. Transparency ensures that the internal mechanisms of AI models are clear and understandable, thereby promoting trust and accountability. Reliability focuses on consistently producing accurate and dependable results, contributing to the overall credibility of AI systems. Causality aims to uncover the cause-and-effect relationships within AI decisions, providing deeper insights into model behaviour. Usability emphasizes making AI systems accessible and user-friendly, catering to diverse stakeholders. Privacy concerns involve safeguarding sensitive information and ensuring responsible data handling practices. Trust is built through transparent processes, reliable outcomes, and ethical considerations. Fairness addresses the need for unbiased and equitable AI decisions, mitigating the impact of algorithmic biases, and thereby promoting ethical and responsible AI practices across various domains.\n

\n 2.1\n Techniques and Methods to Make AI Models More Interpretable and Understandable to Humans\n

\n Weber et al. [\n \n 163\n \n ,\n \n 164\n \n ] highlighted the importance of making AI models more interpretable and understandable to humans for building trust and ensuring responsible deployment of AI systems. AI practitioners can employ certain techniques and methods to enhance model interpretability and understandability, fostering responsible AI deployment and increasing user trust. Table\n \n 1\n \n presents some of these techniques and methods.\n

\n \n Table\u00a01 Techniques, description, strengths, and weaknesses for making AI interpretable and understandable\n \n

\n \n Full size table\n \n \n

\n 2.2\n Understandability of Artificial Intelligence Components\n

\n Artificial Intelligence (AI) is a diverse and multidisciplinary field encompassing numerous elements, methodologies, and specialized areas. These interconnected AI components often collaborate within AI systems to tackle intricate challenges and execute tasks [\n \n 70\n \n ]. The realm of AI research and development continually progresses, yielding novel techniques and breakthroughs across its various domains. In studies by Arrieta et al. [\n \n 10\n \n ] and Montavon et al. [\n \n 108\n \n ], \"understandability\" refers to both the extent to which a human can comprehend a model\u2019s decision and the model\u2019s capacity to elucidate its function to a human. AI comprises several key components that synergize to empower machines to undertake tasks typically requiring human intelligence. These AI components undertake diverse tasks, spanning from natural language comprehension and generation to image object recognition and autonomous decision-making. By harnessing these capabilities, AI systems enhance human intelligence, automate repetitive tasks, and tackle complex challenges across diverse fields. Table\n \n 2\n \n elucidates some of these key components.\n

\n \n Table\u00a02 AI components, description, strength, weakness, and features of AI components\n \n

\n \n Full size table\n \n \n

\n 3\n Application of AI Explainable in Local Use Cases\n

\n AI explainability has many local use cases across different industries and applications [\n \n 15\n \n ]. The local use cases include healthcare, finance, retail and e-commerce, agriculture, education, transport and logistics, and legal and regulatory. These are succinctly presented in the Fig.\n \n 2\n \n and discussed as follows:\n

\n \n Fig.\u00a02\n \n

\n Local use cases of explainable AI\n

\n \n Full size image\n \n \n

\n (i) Healthcare: Explainable Artificial Intelligence (XAI) plays a crucial role in disease diagnosis, AI models used in medical diagnosis can explain their decisions, helping doctors understand the reasoning behind a diagnosis. Ensuring explainability in AI for disease diagnosis is crucial for fostering trust, transparency, and acceptance of AI-based diagnostic systems within the healthcare sector. As highlighted in study by Asan et al. [\n \n 11\n \n ], when patients and healthcare providers comprehend the reasoning behind AI-generated diagnoses, they can make more informed decisions and place greater trust in the recommendations. Incorporating explainability into AI disease diagnosis not only assists healthcare professionals but also addresses concerns regarding safety, accountability, and ethical utilization of AI in healthcare [\n \n 21\n \n ]. It is imperative to strike a balance between model accuracy and interpretability to establish trust in the AI system and enable its effective integration into medical practice. According to Zhang et al. [\n \n 170\n \n ], eXplainable AI in healthcare focuses on several critical aspects, including ensuring transparent data and features, thereby enabling healthcare providers to comprehend the data inputs and variables influencing the diagnostic process. Secondly, XAI provides insights into feature importance and contribution, enabling clinicians to identify critical factors contributing to a diagnosis. Additionally, it offers visualization tools and graphs to represent complex data relationships visually, aiding in comprehension and decision-making. Moreover, XAI enhances model interpretability, allowing healthcare providers to understand how a diagnostic model arrives at its conclusions. Furthermore, Schoonderwoerd et al. [\n \n 140\n \n ] explained that XAI facilitates natural language explanations, which help communicate diagnostic results and treatment recommendations in a comprehensible manner to both clinicians and patients. XAI also supports treatment recommendations, comparative analysis between different diagnostic approaches, and patient education, thereby empowering healthcare professionals and patients with actionable insights and knowledge for informed decision-making in disease diagnosis and management. For instance, a recent study by Zhu [\n \n 171\n \n ] utilized computational intelligence with AI to cclassify brain images from MRI or CT scans, demonstrating the impact of dosage ratios on health over treatment duration, and diagnosing memory impairment in psychoactive substance users. The shift towards digital pathology allows pathologists to utilize AI-based computer programs for analyzing whole slide images, but many top-performing algorithms operate as \"black boxes,\" leaving even developers uncertain about their decision-making processes [\n \n 120\n \n ]. This review emphasizes the necessity of explainability in digital pathology to prevent mistakes and ensure patient safety, introducing various explainable AI techniques that enhance the transparency of these algorithms. Furthermore, it advocates for the development of novel user interfaces that promote contextual understanding and interactive exploration, enabling medical experts to engage with AI results more effectively and maintain a critical human presence in the decision-making process.\n

\n (ii) Finance: Weber et al. [\n \n 163\n \n ,\n \n 164\n \n ], the explainability of AI in finance, particularly in credit scoring and risk assessment, is crucial for several reasons, including regulatory compliance, transparency, and trust-building with both lenders and borrowers. In credit scoring, AI models can explain why an applicant\u2019s credit score is calculated in a certain way, offering transparency to applicants and regulators while in risk assessment, explainable AI models can provide insights into the factors influencing risk assessment in financial investments. Explainability in AI for credit scoring and risk assessment helps lenders make responsible lending decisions and empowers borrowers to understand and improve their financial profiles [\n \n 62\n \n ]. Moreover, it assists regulatory bodies in ensuring that AI-driven credit decisions comply with legal and ethical standards, promoting fairness and accountability in the financial industry. Matcov [\n \n 103\n \n ] explained that XAI is instrumental in credit scoring and risk assessment within the finance sector, operating across several key dimensions like identifying feature importance, and allowing financial institutions to understand the factors driving creditworthiness and risk. Talaat et al. [\n \n 146\n \n ] showed that XAI also provides a breakdown of credit scores, offering transparency into how individual scores are calculated and the factors influencing them. Moreover, it enables comparative analysis, allowing institutions to compare different models or approaches to credit scoring and risk assessment to determine the most effective strategies. Heng and Subramanian [\n \n 62\n \n ] presented that XAI further identifies key risk factors associated with credit decisions, aiding in the identification and mitigation of potential risks. The concept of AI has helped to gain insight into the marketplace, for example, Liu et al. [\n \n 94\n \n ,\n \n 96\n \n ] pinpointed that 279.87 million US internet job posts between 2010 and 2022, the demand for statistical talent with expertise in AI increased by an astounding 31 times, resulting in 932 different AI-related employment categories. This insight assists in enhancing the recruitment and employability of job seekers. Additionally, in the work of Fritz-Morgenthal, et al. [\n \n 48\n \n ], XAI ensures regulatory compliance by providing explanations and justifications for credit decisions, helping financial institutions adhere to regulatory requirements and ethical standards while making informed lending decisions. Overall, XAI enhances transparency, accountability, and effectiveness in credit scoring and risk assessment processes, enabling financial institutions to make more accurate and fair credit decisions while ensuring regulatory compliance. The banking industry faces challenges in loan acceptance, with rates dropping to 61\u201370% and further decreasing to 50% post-pandemic, leaving many applicants unaware of the reasons for their rejections, which negatively impacts their credit scores. To address this issue, Nallakaruppan et al. [\n \n 110\n \n ] proposed an XAI framework integrated with Random Forest algorithms, offering clear explanations for loan decisions, thus enhancing transparency in the lending process. This XAI framework utilizes LIME and SHAP explainers to provide local and global insights on various parameters, significantly improving the accuracy and reliability of loan acceptance predictions.\n

\n (iii) Retail and e-commerce: Explainability is gaining increasing significance in retail and e-commerce across various applications, from tailored product suggestions to fraud detection. In the fast-paced world of technology, companies in e-commerce are leveraging AI through applications like recommendation systems, fraud detection, and fake filters to enhance competitiveness. However, a significant challenge remains in understanding and explaining the outcomes produced by AI algorithms, which impacts their trustworthiness and raises ethical and privacy concerns. In e-commerce, where timely decisions are essential, integrating XAI systems is vital. Chaudhary et al. [\n \n 27\n \n ] worked systems that are designed to deliver immediate justifications, addressing the shortcomings of existing technologies that often fail to provide comprehensive explanations of the decision-making processes underlying AI-generated results or recommendations. The implementation of XAI not only fosters customer trust and satisfaction but also equips retailers to make informed decisions and streamline their operations [\n \n 13\n \n ]. It enables retailers to harmonize personalized experiences with user privacy and ethical considerations. By elucidating the rationale behind individual recommendations or decisions, explainable AI enhances the speed and accuracy of decision-making processes by complementing consumers\u2019 interactions with AI systems. Leveraging customer data, explainable AI facilitates decision-making, leading to improved business outcomes. This transparency and reasoning contribute to reducing abandoned products, increasing order value, and consequently boosting revenue and conversion rates, as noted [\n \n 78\n \n ]. One prominent application of AI in e-commerce is the customization of product recommendations, leveraging insights from prior purchases, searches, and browsing behaviors to enhance average order value. Moreover, AI-driven inventory management leverages historical sales data, predicted demand shifts, and potential supply chain issues to optimize inventory levels, unlike traditional methods limited to current stock levels. Additionally, online retailers are adopting chatbots and digital assistants to offer round-the-clock customer support, as highlighted [\n \n 60\n \n ].\n

\n (iv) Agriculture: Sabrina et al. [\n \n 134\n \n ] state that AI explainability plays a vital role in the agriculture industry, helping farmers, agronomists, and policymakers understand and trust AI-driven recommendations and decisions. Explainability in agriculture supports sustainable farming practices, optimal resource utilization, and resilient crop management. It also helps agricultural stakeholders to make informed decisions, improve yields, reduce resource waste, and address ethical and regulatory considerations while fostering trust in AI-driven solutions. The deployment of XAI accounts for the experience and authority of the farmers and comprehension increases the farmer\u2019s trust in AI systems and makes them more effective [\n \n 64\n \n ]. Farmers can explore the knowledge acquired by machine learning models trained to identify suitable crops for cultivation through eXplainable AI (XAI) methods and visualizations. With XAI charts, even individuals without expertise in machine learning can comprehend the rationale behind a model\u2019s prediction for a specific crop in a given scenario [\n \n 23\n \n ]. Leveraging ML applications like computer vision and time-series analysis, farmers can anticipate the ideal timing for seeding, harvesting, and marketing their crops, thereby enhancing their decision-making capabilities [\n \n 50\n \n ]. Effective decision-making processes leading to high and high-quality crop yields involve a comprehensive understanding of weather patterns, along with the use of crop prediction methodologies that consider various soil parameters and atmospheric conditions [\n \n 41\n \n ]. XAI plays a transformative role in agriculture by empowering farmers to make informed decisions through transparent insights into crop health, growth patterns, and nutrient requirements derived from data-driven analyses. Moreover, Linheiro, et al. [\n \n 91\n \n ] work shows that XAI enhances weather forecasting accuracy by integrating complex weather data with historical patterns, allowing farmers to anticipate and mitigate the impact of adverse weather conditions on crops. In precision agriculture, XAI optimizes resource allocation by offering detailed insights into soil conditions, moisture levels, and crop health at a granular level, facilitating targeted interventions such as irrigation and fertilization. Mana et al. [\n \n 102\n \n ], explain that XAI aids in harvest prediction by analyzing historical data, crop maturity levels, and environmental factors to forecast optimal harvest times, maximizing yield, and minimizing waste. Furthermore, XAI contributes to disease and pest detection by leveraging machine learning algorithms to identify early signs of crop diseases and pest infestations, enabling proactive measures such as pest control and crop rotation to protect crop yields and ensure food security. In the study of Chang-Brahim, et al. [\n \n 26\n \n ], Marker-assisted selection (MAS) significantly enhances crop breeding by enabling the quick identification of plants with desired traits, with precise phenotyping being essential for its effectiveness. Recent advancements in high-throughput remote phenotyping using unmanned aerial vehicles and machine learning offer efficient alternatives to traditional methods, while AI and explainable AI address the limitations of genome-wide association studies (GWAS) in understanding complex traits like drought tolerance. This review outlines the state-of-the-art methods and processes in MAS, highlighting the integration of explainable AI throughout the breeding workflow from phenotyping to association analyses. Overall, XAI empowers farmers with actionable insights and decision support tools to optimize crop productivity, improve resilience against environmental challenges, and enhance sustainability in agriculture practices.\n

\n (v) Education: Explainable Artificial Intelligence plays a crucial role in elucidating the insights uncovered by AI systems, which can unveil novel patterns and discoveries from data, thereby opening up new avenues for research across various disciplines. XAI has the potential to revolutionize the way students perceive, absorb, and comprehend course materials, facilitating enhanced knowledge transfer and optimization of classroom learning experiences. Moreover, XAI can provide valuable insights to educators in learning analytics, highlighting areas where students may require additional support and targeted intervention as well as reshaping the lack of knowledge-sharing system [\n \n 125\n \n ,\n \n 128\n \n ]. By promoting transparency and accountability, XAI assists educational institutions and service providers in addressing trust issues associated with AI usage, thereby unlocking its potential benefits [\n \n 77\n \n ]. XAI is reshaping education across various facets, starting with adaptive learning, where it tailors instructional content and pace to individual student needs based on comprehensive data analysis, fostering personalized learning experiences and improving student outcomes. In assessment and grading, XAI enhances objectivity and fairness by providing transparent insights into grading criteria and performance metrics, enabling educators to make informed evaluations and provide constructive feedback [\n \n 67\n \n ,\n \n 69\n \n ]. Moreover, Lee et al. [\n \n 89\n \n ], expressed that XAI empowers educators with learning analytics tools that analyze student data to identify patterns, trends, and areas for improvement, facilitating data-driven decision-making and instructional planning. In resource allocation, XAI optimizes the allocation of educational resources such as personnel, materials, and technology by analyzing student needs, preferences, and performance data, ensuring efficient utilization and equitable access [\n \n 13\n \n ]. Also, XAI combined with federated learning (FL) and multimodal fusion eliminates low-quality training models on the server side, minimizing the detrimental effects of low-quality data, data duplication, and inconsistency reasoning without data transmission for human-centric virtual reality [\n \n 160\n \n ,\n \n 165\n \n ,\n \n 172\n \n ]. As a rising approach to enhancing trust in AI systems, XAI advocates for the development of methods that provide clear explanations and justifications for the decisions made by AI. Liu, et al. [\n \n 94\n \n ,\n \n 96\n \n ] offer an overview of the applications of XAI in education, including examples from empirical studies that address several key questions: What are the primary categories of XAI techniques employed to interpret and clarify AI models? What distinct features characterize the application of XAI in the educational context? How can XAI models assist teachers, students, and researchers in making informed decisions? Additionally, the work explores the challenges and opportunities associated with the use of XAI in education. Additionally, XAI promotes parent and guardian engagement by providing transparent explanations of student progress, learning goals, and educational interventions, fostering collaboration between educators and families to support student success and well-being. Overall, XAI enhances the effectiveness, efficiency, and equity of education by providing transparent insights, personalized support, and data-driven decision-making tools to educators, students, and families alike.\n

\n (vi) Transport and Logistics: Explainable Artificial Intelligence is transforming transportation and logistics by offering transparent insights and facilitating data-driven decision-making across diverse sectors. Automobile traffic accidents [\n \n 166\n \n ] pose a significant threat to public safety, leading to numerous injuries and fatalities each year. Abdulrashid et al. [\n \n 1\n \n ] present a comprehensive design for an explainable artificial intelligence (XAI) artifact that integrates accident data for use by various stakeholders and decision-makers. It introduces responsible and interpretable models along with a taxonomy that categorizes driver-related behaviors linked to different levels of injury severity. In the initial phase, advanced techniques such as Bayesian dynamic conditional imputation for missing data and SHapley Additive exPlanations (SHAP) for risk factor analysis were employed to address data issues and uncover spatiotemporal elements influencing accidents. The final phase includes the classification of crash data into causal factors using aggregate SHAP scores, demonstrating how different risk factors contribute to varying levels of injury severity. The enhanced transparency provided by XAI has the potential to enable traceability, leading to a more precise evaluation of due diligence and liability in the event of accidents, thereby benefiting insurance companies and regulatory bodies. For example, Kuznietsov et al. [\n \n 85\n \n ] highlighted the advantages of XAI in the inspection, debugging, and auditing of AI models, which can enhance their robustness and elevate trust levels in autonomous vehicles. XAI plays a crucial role in reinstating the contestability of decisions made by autonomous driving systems, potentially safeguarding users from undue harm and establishing a basis for legal recourse. Mahbooba et al. [\n \n 100\n \n ] underscored the necessity of XAI in Intelligent Connected Vehicles (ICV) for reducing vulnerability and detecting intrusions, emphasizing the importance of explanation and interpretability of classification model decisions, as well as effectively managing evolving threats and attack mechanisms. XAI also fosters the development of machine learning algorithms that employ more explainable techniques while maintaining efficient learning performance. Furthermore, it facilitates human comprehension of model actions and decisions, aiding in information extraction and the visualization of results with maximal transparency [\n \n 115\n \n ]. In route planning and optimization, XAI analyses historical data, wireless enabled self-powered sensing for logistics in an industry, traffic patterns, and logistical constraints to recommend efficient routes and scheduling, minimizing costs and maximizing delivery efficiency [\n \n 93\n \n ]. XAI enhances demand forecasting accuracy by analyzing market trends, customer behaviour, and historical data to predict future demand, enabling companies to optimize inventory management and resource allocation. In freight load optimization, XAI leverages predictive analytics and optimization algorithms to maximize cargo capacity utilization, reducing empty miles and fuel consumption while increasing profitability. In the work El Makhloufi [\n \n 40\n \n ], XAI enables real-time tracking of shipments and vehicles, providing stakeholders with visibility into the supply chain and enabling proactive interventions to address delays or disruptions. Furthermore, XAI helps mitigate the environmental impact of transportation by optimizing route planning, load distribution, and vehicle utilization to minimize emissions and carbon footprint, promoting sustainable and eco-friendly practices in the transportation and logistics industry (T. [\n \n 18\n \n ,\n \n 95\n \n ]). Overall, XAI empowers transportation and logistics companies with transparent insights and data-driven solutions to optimize operations, improve efficiency, and reduce environmental impact across the supply chain.\n

\n (vii) Legal and Regulatory: In legal and regulatory contexts, Explainable AI is applied to clarify and justify AI model decisions, ensuring alignment with legal and ethical standards. XAI aids regulators in understanding complex AI algorithms, making it easier to detect and prevent biases, discrimination, or unfair practices in areas like loan approvals, hiring, and sentencing. By enhancing transparency, XAI fosters accountability, supports audits, and helps organizations meet compliance obligations, such as the GDPR in the EU, which mandates the right to explanations for automated decisions. Fresz et al. [\n \n 47\n \n ] paper examines the intersection of law and eXplainable Artificial Intelligence, noting that while the AI Act adopted by the European Parliament in March 2024 is widely discussed, other legal areas remain underexplored. Focusing primarily on European and, to some extent, German law, it also considers international concepts such as fiduciary duties, the General Data Protection Regulation (GDPR), and product safety and liability. By analysing XAI taxonomies, the paper identifies specific requirements for XAI methods in various legal domains, concluding that each area necessitates different properties from XAI. It highlights that the current advancements in XAI do not adequately meet these legal requirements, particularly concerning the correctness (or fidelity) and confidence estimates of XAI methods. This transparency builds trust in AI systems and ensures they adhere to ethical and legal frameworks. Kesari et al. [\n \n 76\n \n ], article address the legal need for explainable AI (XAI) in automated decision-making, which often lacks transparency, posing risks to accountability and the public trust which introduces a Legal-XAI framework that categorizes explanations for AI decisions in law, distinguishing between types like global vs. local and comprehensive vs. selective explanations. The authors explore how legal, technical, and behavioral factors influence the choice of explanation method, using credit scoring as a practical example. They advocate for policymakers to specify explanation requirements to align with their objectives, making automated decisions more understandable and democratic. The article bridges gaps between legal and computer science fields, offering a roadmap and software to compare algorithmic explanation methods through empirical field experiments. Mueller et al., (2022), presented that AI is rapidly advancing in healthcare, especially in the field of in vitro diagnostics, but the new European regulation on In Vitro Diagnostic Medical Devices (IVDR) requires AI software in medical devices to be explainable. This requirement challenges the use of machine learning and deep learning in these devices, as they often lack necessary explanatory components. Trustworthy AI is needed to enable healthcare professionals to make informed, accountable decisions, which makes explainable AI essential. Techniques like layer-wise relevance propagation can help visualize which parts of the input data influenced a neural network\u2019s decision. This paper introduces \"causability,\" a concept for measuring the quality of AI explanations, which alongside explainability, is crucial for validating the scientific, analytical, and clinical performance of AI-based diagnostic tools. Tax administrations worldwide are increasingly utilizing artificial intelligence (AI) systems for automation. However, this automation significantly impacts the rights of taxpayers who are subject to algorithmic assessments, a situation exacerbated by the complexity and opacity of these AI systems. Ku\u017aniacki et al. [\n \n 84\n \n ] article contends that protecting taxpayers\u2019 rights requires the implementation of XAI technologies, which can make the operations and decisions of tax-related AI systems comprehensible to taxpayers, administrative appeal bodies, and the courts. This need arises from constitutional principles governing taxation but remains inadequately addressed by both soft and hard law frameworks, which fail to meet the specific informational needs of the tax sector. To fill this gap, the authors concluded by identifying the technical and legal challenges associated with effectively applying explanation techniques to tax AI, ensuring that automation does not compromise taxpayers\u2019 rights.\n

\n 4\n Challenges and Solutions of AI Explainable in Local Use Cases\n

\n As Artificial Intelligence (AI) continues to permeate various sectors and industries, the demand for transparency and interpretability in AI systems has become increasingly pronounced. In local use cases, where AI is applied within specific contexts such as healthcare, finance, retail, agriculture, education, transport, and logistics, ensuring explainability presents unique challenges and opportunities. Work by D\u00edaz-Rodr\u00edguez, et al. [\n \n 36\n \n ] discussed the challenges and solutions associated with AI explainability in local use cases, highlighting the importance of understanding how AI systems arrive at their decisions to foster trust, accountability, and ethical AI deployment discussed in this section. By addressing these challenges with tailored solutions, stakeholders can harness the full potential of AI while mitigating risks and ensuring responsible and impactful AI applications in diverse local contexts as shown in Table\n \n 3\n \n .\n

\n \n Table\u00a03 Local use cases, challenges, and solutions of XAI\n \n

\n \n Full size table\n \n \n

\n 5\n Open Research Directions\n

\n Some of the identified areas for further research are highlighted below:\n

\n (i) Integration of Human Feedback: One promising research direction involves the integration of human feedback into explainable AI systems to enhance interpretability and understanding. Saeed and Omlin [\n \n 136\n \n ] extracted that current XAI techniques aim to provide transparent explanations for AI decisions, incorporating user feedback can further refine these explanations and improve their relevance and usefulness. Research in this area could explore methods for soliciting and incorporating feedback from end-users, domain experts, and stakeholders to iteratively improve the interpretability of AI models. This could involve developing interactive XAI interfaces that allow users to query the system, provide feedback on generated explanations, and contribute domain-specific knowledge to enhance model interpretability.\n

\n (ii) Explainability in Deep Learning Models: Another crucial area of research involves the advancement of explainability methods tailored for deep learning models [\n \n 166\n \n ], which often function as opaque entities due to their intricate architectures and high-dimensional feature representations. While recent progress in eXplainable AI (XAI) has made significant strides in interpreting simpler models like decision trees or linear models, deep learning models present distinct challenges for explainability. Future research endeavors could concentrate on devising innovative techniques that offer transparent explanations for deep learning models, including attention mechanisms, convolutional neural networks (CNNs), and recurrent neural networks (RNNs) [\n \n 66\n \n ]. This may entail the exploration of approaches for visualizing and interpreting the internal representations of deep learning models, extracting pertinent features, and elucidating the decision-making processes of intricate neural networks.\n

\n (iii) Contextual Explainability: Furthermore, research into contextual explainability is gaining traction, especially for local use cases where AI systems function within specific domains or environments. Contextual explainability involves AI systems offering explanations that are relevant and tailored to the specific context or user preferences. This requires considering factors such as cultural norms, regulatory requirements, and user expectations when generating explanations for AI decisions [\n \n 33\n \n ]. Nagahisarchoghaei et al. [\n \n 109\n \n ] suggested that future research could focus on creating context-aware XAI techniques that adapt explanations to the distinct characteristics of various local use cases, ensuring that AI systems deliver interpretable and actionable insights that are meaningful and relevant to users.\n

\n (iv) Ethical and Social Implications: Future research in XAI should address the ethical and social implications of explainable AI systems, particularly in local use cases where AI applications can significantly impact society. This includes tackling issues such as fairness, transparency, accountability, and privacy in AI decision-making processes. Researchers could investigate methods for detecting and mitigating biases in XAI systems to ensure explanations are accurate, unbiased, and trustworthy. Additionally, it is important to examine the broader societal implications of XAI, including its effects on job displacement, human-AI collaboration, and the distribution of power and resources in society [\n \n 139\n \n ]. By considering these ethical and social factors, researchers can ensure that XAI technologies are developed and deployed responsibly, with careful consideration of their potential impacts on individuals, communities, and society at large.\n

\n 6\n Conclusion\n

\n The recent advancements in Explainable Artificial Intelligence (XAI) represent a crucial step toward bridging the gap between complex AI models and human understanding. By improving explanatory depth, XAI techniques provide users with insights into AI decision-making processes, thereby enhancing transparency and trust. As AI continues to evolve and integrate into various societal sectors, the development of effective explanation methods remains a priority in AI research and development. These explanations aim to make AI system behavior transparent and interpretable to humans, fostering trust, accountability, and informed decision-making across various applications. As demonstrated in this work, developers, operators, and users generally expect XAI to clarify key questions, enabling people to fully understand and trust AI system decisions. This trust will allow users to have confidence in the models and encompass all XAI components related to transparency, causality, bias, fairness, and safety. However, the lack of explainability and trust currently limits the full utilization of AI systems.\n

\n Furthermore, the implementation of XAI in local use cases has great potential to improve decision-making processes in various fields such as healthcare, finance, and law. However, successful utilization requires careful consideration of local data characteristics, regulatory requirements, and user preferences. As XAI continues to evolve, several key areas of future research will shape its development. These include enhancing model transparency by devising methods to make complex AI models, like deep learning models, more transparent and interpretable without significantly compromising their performance. This research offers practical implications, for instance, facilitating better decision-making processes in industries by providing transparent and interpretable insights into AI models. Also, implementing emerging techniques in XAI can foster trust and acceptance among users, as they gain a deeper understanding of AI model behaviour and reasoning, leading to increased confidence in the technology. Furthermore, the study emphasizes the importance of collaboration between computer scientists, ethicists, psychologists, sociologists, and other disciplines, highlighting the need for a holistic approach to address the multifaceted challenges of XAI effectively.\n

\n Data and Material Availability\n

\n No datasets were generated or analysed during the current study.\n

\n Code Availability\n

\n Not applicable.\n

\n References\n

\n
\n Abdulrashid I, Farahani RZ, Mammadov S, Khalafalla M, Chiang WC (2024) Explainable artificial intelligence in transport logistics: risk analysis for road accidents. Transp Res Part E: Logist Transp Rev 186:103563\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Adadi A, Berrada M (2018) Peeking inside the black box: a survey on explainable artificial intelligence (XAI). IEEE Access 6:52138\u201352160\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Alicioglu G, Sun B (2022) A survey of visual analytics for explainable artificial intelligence methods. Comput Graph 102:502\u2013520\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Aloisi C (2023) The future of standardised assessment: validity and trust in algorithms for assessment and scoring. Eur J Educ 58(1):98\u2013110\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Amann J, Blasimme A, Vayena E, Frey D, Madai VI,Precise4Q Consortium (2020) Explainability for artificial intelligence in healthcare: a multidisciplinary perspective. BMC Med Inf Decis Mak, 20, pp 1\u20139\n
\n
\n
\n Andreu-Perez J, Deligianni F, Ravi D, Yang GZ (2018) Artificial Intelligence and Robotics. arXiv preprint\n \n arXiv:1803.10813\n \n .\n
\n
\n
\n Anica-Popa I, Anica-Popa L, R\u0103dulescu C, Vr\u00eencianu M (2021) The integration of artificial intelligence in retail: benefits, challenges and a dedicated conceptual framework. Amfiteatru Economic 23(56):120\u2013136\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Antoniadi AM, Du Y, Guendouz Y, Wei L, Mazo C, Becker BA, Mooney C (2021) Current challenges and future opportunities for XAI in machine learning-based clinical decision support systems: a systematic review. Appl Sci 11(11):5088\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Araujo H, Mousavi MR, Varshosaz M (2023) Testing, validation, and verification of robotic and autonomous systems: a systematic review. ACM Transact Softw Eng Methodol 32(2):1\u201361\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Arrieta AB, D\u00edaz-Rodr\u00edguez N, Del Ser J, Bennetot A, Tabik S, Barbado A, Herrera F (2020) Explainable Artificial Intelligence (XAI): concepts, taxonomies, opportunities, and challenges toward responsible AI. Inf Fus 58:82\u2013115\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Asan O, Bayrak AE, Choudhury A (2020) Artificial intelligence and human trust in healthcare: focus on clinicians. J Med Internet Res 22(6):e15154\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Bahrammirzaee A (2010) A comparative survey of artificial intelligence applications in finance: artificial neural networks, expert system, and hybrid intelligent systems. Neural Comput Applic 19:1165\u20131195.\n \n https://doi.org/10.1007/s00521-010-0362-z\n \n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Belghachi M (2023) A review of explainable artificial intelligence methods, applications, and challenges. Indones J Electr Eng Inf (IJEEI) 11(4):1007\u20131024\n
\n
\n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Bench-Capon TJ (2014) Knowledge representation: an approach to artificial intelligence (vol 32). Elsevier, Publisher, pp 220\n
\n
\n
\n Bharati S, Mondal MRH, Podder P (2023) A review on explainable artificial intelligence for healthcare: Why, How, and When? IEEE Transact Artif Intell 5(4):1429\u20131442.\n \n https://doi.org/10.1109/TAI.2023.3266418\n \n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Bhatt U, Xiang A, Sharma S, Weller A, Taly A, Jia Y, Ghosh J, Puri R, Moura JM, Eckersley P (2020) Explainable machine learning in deployment. In: Proceedings of the 2020 conference on fairness, accountability, and transparency (pp 648\u2013657).\n
\n
\n
\n Barria-Pineda J (2020). Exploring the need for transparency in educational recommender systems. In: Proceedings of the 28th ACM conference on user modeling, adaptation and personalization (pp 376\u2013379).\n
\n
\n
\n Bibri SE, Krogstie J, Kaboli A, Alahi A (2024) Smarter eco-cities and their leading-edge artificial intelligence of things solutions for environmental sustainability: a comprehensive systematic review. Environ Sci Ecotechnology 19:100330\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Bobek S, Kuk M, Szel\u0105\u017cek M, Nalepa GJ (2022) Enhancing cluster analysis with explainable AI and multidimensional cluster prototypes. IEEE Access 10:101556\u2013101574\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Borgonovo E, Plischke E (2016) Sensitivity analysis: a review of recent advances. Eur J Oper Res 248(3):869\u2013887\n
\n
\n \n Article\n \n \n MathSciNet\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Borys K, Schmitt YA, Nauta M, Seifert C, Kr\u00e4mer N, Friedrich CM, Nensa F (2023) Explainable AI in medical imaging: an overview for clinical practitioners\u2013beyond saliency-based XAI approaches. Eur J Radiol 162:110786\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Burzykowski T, Rousseau AJ, Geubbelmans M, Valkenborg D (2023) Introduction to machine learning. Am J Orthod Dentofac Orthop 163(5):732\u2013734\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Cartolano A, Cuzzocrea A, Pilato G (2024). Agriculture Environ.\n \n https://doi.org/10.1007/s11042-023-17978-z\n \n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Chakraborty S, Tomsett R, Raghavendra R, Harborne D, Alzantot M, Cerutti F, Gurram, P (2017) Interpretability of deep learning models: a survey of results. In 2017 IEEE smartworld, ubiquitous intelligence & computing, advanced & trusted computed, scalable computing & communications, cloud & big data computing, Internet of people and smart city innovation (smartworld/SCALCOM/UIC/ATC/CBDcom/IOP/SCI) (pp 1\u20136). IEEE.\n
\n
\n
\n Chamola V, Hassija V, Sulthana AR, Ghosh D, Dhingra D, Sikdar B (2023) A review of trustworthy and explainable artificial intelligence (xai). IEEE Access.\n
\n
\n
\n Chang-Brahim I, Koppensteiner LJ, Beltrame L, Bodner G, Saranti A, Salzinger J, Molin EM (2024) Reviewing the essential roles of remote phenotyping, GWAS and explainable AI in practical marker-assisted selection for drought-tolerant winter wheat breeding. Front Plant Sci 15:1319938\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Chaudhary M, Gaur L, Singh G, Afaq A (2024) Introduction to explainable AI (XAI) in e-commerce. In: Role of explainable artificial intelligence in e-commerce (pp 1\u201315). Cham: Springer Nature Switzerland.\n
\n
\n
\n Choo J, Liu S (2018) Visual analytics for explainable deep learning. IEEE Comput Graphics Appl 38(4):84\u201392\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Chowdhary K, Chowdhary KR (2020). Natural language processing. Fundamentals of artificial intelligence, 603\u2013649.\n
\n
\n
\n Chromik M, Eiband M, Buchner F, Kr\u00fcger A, Butz A (2021) I think I get your point, AI! The illusion of explanatory depth in explainable AI. In: 26th International Conference on Intelligent User Interfaces (pp. 307\u2013317).\n
\n
\n
\n Cohen SN, Snow D, Szpruch L (2021) Black-box model risk in finance. arXiv preprint\n \n arXiv:2102.04757\n \n , 10.\n
\n
\n
\n Cortez P, Embrechts MJ (2013) Using sensitivity analysis and visualization techniques to open black box data mining models. Inf Sci 225:1\u201317\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Dazeley R, Vamplew P, Foale C, Young C, Aryal S, Cruz F (2021) Levels of explainable artificial intelligence for human-aligned conversational explanations. Artif Intell 299:103525\n
\n
\n \n Article\n \n \n MathSciNet\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Del Ser J, Barredo-Arrieta A, D\u00edaz-Rodr\u00edguez N, Herrera F, Saranti A, Holzinger A (2024) On generating trustworthy counterfactual explanations. Inf Sci 655:119898\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Deldjoo Y, Nazary F, Ramisa A, Mcauley J, Pellegrini G, Bellogin A, Noia TD (2023) A review of modern fashion recommender systems. ACM Comput Surv 56(4):1\u201337\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n D\u00edaz-Rodr\u00edguez N, Del Ser J, Coeckelbergh M, de Prado ML, Herrera-Viedma E, Herrera F (2023) Connecting the dots in trustworthy artificial intelligence: from AI principles, ethics, and key requirements to responsible AI systems and regulation. Inf Fus 99:101896\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Dwivedi R, Dave D, Naik H, Singhal S, Omer R, Patel P, Ranjan R (2023) Explainable AI (XAI): core ideas, techniques, and solutions. ACM Comput Surv 55(9):1\u201333\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Eberhard K (2023) The effects of visualization on judgment and decision-making: a systematic literature review. Manage Rev Quart 73(1):167\u2013214.\n \n https://doi.org/10.1007/s11301-021-00235-8\n \n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Ehsan U, Passi S, Liao QV, Chan L, Lee IH, Muller M, Riedl MO (2024) The who in XAI: how AI background shapes perceptions of AI explanations. In: Proceedings of the CHI conference on human factors in computing systems (pp 1\u201332).\n
\n
\n
\n El Makhloufi A (2023) AI application in transport and logistics: opportunities and challenges (An Exploratory Study).\n
\n
\n
\n Eli-chukwu N (2020) Applications of Artificial Intelligence in Agriculture\u202f: A Review. July.\n \n https://doi.org/10.48084/etasr.2756\n \n
\n
\n
\n Elton DC (2020) Self-explaining AI as an alternative to interpretable AI. In: artificial general intelligence: 13th international conference, AGI 2020, St. Petersburg, Russia, September 16\u201319, 2020, Proceedings 13 (pp. 95\u2013106). Springer International Publishing\n
\n
\n
\n Ennab M, Mcheick H (2022) Designing an interpretability-based model to explain the artificial intelligence algorithms in healthcare. Diagnostics 12(7):1557\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Fanni SC, Febi M, Aghakhanyan G, Neri E (2023) Natural language processing. Introduction to artificial intelligence. Springer International Publishing, Cham, pp 87\u201399\n
\n
\n \n Chapter\n \n \n Google Scholar\n \n
\n
\n
\n Felzmann H, Fosch-Villaronga E, Lutz C, Tam\u00f2-Larrieux A (2020) Towards transparency by design for artificial intelligence. Sci Eng Ethics 26(6):3333\u20133361\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Fern\u00e1ndez Llorca D, G\u00f3mez E (2021) Trustworthy autonomous vehicles. Publications Office of the European Union, Luxembourg, p 30942\n
\n
\n \n Google Scholar\n \n
\n
\n
\n Fresz B, Dubovitskaya E, Brajovic D, Huber MF, Horz C (2024) How should AI decisions be explained? Requirements for explanations from the perspective of European Law. In: Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society (Vol. 7, pp. 438\u2013450).\n
\n
\n
\n Fritz-Morgenthal S, Hein B, Papenbrock J (2022) Financial risk management and explainable, trustworthy, responsible AI. Front Artif Intell 5:779799\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Fuhrman JD, Gorre N, Hu Q, Li H, El Naqa I, Giger ML (2022) A review of explainable and interpretable AI with applications in COVID-19 imaging. Med Phys 49(1):1\u201314\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Gardezi M, Joshi B, Rizzo DM, Ryan M, Prutzer E, Brugler S, Dadkhah A, Ryan M (2023). Artificial intelligence in farming\u202f: Challenges and opportunities for building trust opportunities for improving. March, 1\u201312.\n \n https://doi.org/10.1002/agj2.21353\n \n
\n
\n
\n Ge Y, Liu S, Fu Z, Tan J, Li Z, Xu S, Li Y, Xian Y, Zhang Y (2022) A survey on trustworthy recommender systems. arXiv preprint\n \n arXiv:2207.12515\n \n .\n
\n
\n
\n Gelfond M, Kahl Y (2014). Knowledge representation, reasoning, and the design of intelligent agents: The answer-set programming approach. Cambridge University Press.\n
\n
\n
\n Gerke S, Minssen T, Cohen G (2020) Ethical and legal challenges of artificial intelligence-driven healthcare. In Artificial intelligence in healthcare (pp 295\u2013336). Academic Press.\n
\n
\n
\n Gong J, Yu Q, Li T, Liu H, Zhang J, Fan H, Jin D, Li Y\u00a0(2023). Demo: scalable digital twin system for mobile networks with generative AI. In: Proceedings of the 21st annual international conference on mobile systems, applications and services (MobiSys\u2019 23), Association for computing machinery.\n \n https://doi.org/10.1145/3581791.3597297\n \n .\n
\n
\n
\n Govers FX (2018) Artificial intelligence for robotics: Build intelligent robots that perform human tasks using AI techniques. Packt Publishing Ltd.\n
\n
\n
\n Grm K, \u0160truc V, Artiges A, Caron M, Ekenel HK (2018) Strengths and weaknesses of deep learning models for face recognition against image degradations. It Biometrics 7(1):81\u201389\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Guidotti R, Monreale A, Ruggieri S, Pedreschi D, Turini F, Giannotti F (2018). Local rule-based explanations of black box decision systems. arXiv preprint\n \n arXiv:1805.10820\n \n .\n
\n
\n
\n Hagras H (2018) Toward human-understandable, explainable AI. Computer 51(9):28\u201336\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Hanif A, Zhang X, Wood S (2021) A survey on explainable artificial intelligence techniques and challenges. In: 2021 IEEE 25th international enterprise distributed object computing workshop (EDOCW) (pp. 81\u201389). IEEE.\n
\n
\n
\n Hanif PM, Soni M, Soni A (2023) A study on the role of artificial intelligence in e-commerce. J Emerg Technol Innov Res 10(9):105\u2013110\n
\n
\n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Hashemi M (2023) Who wants what and how: a mapping function for explainable artificial intelligence. arXiv preprint\n \n arXiv:2302.03180\n \n .\n
\n
\n
\n Heng YS, Subramanian P (2022) A systematic review of machine learning and explainable artificial intelligence (XAI) in credit risk modelling. In: Proceedings of the future technologies conference (pp 596\u2013614). Cham: Springer International Publishing.\n
\n
\n
\n Holstein K, Doroudi S (2022) Equity and artificial intelligence in education. In: The ethics of artificial intelligence in education (pp 151\u2013173) Routledge.\n
\n
\n
\n H\u00fcllmann JA (2022) Explainable AI in farming\u202f: configurations of human-AI joint decision-making. 1\u20134.\n
\n
\n
\n Hulsen T (2023) Explainable artificial intelligence (XAI) in healthcare.\n
\n
\n
\n Ibrahim R, Shafiq MO (2023) Explainable convolutional neural networks: a taxonomy, review, and future directions. ACM Comput Surv 55(10):1\u201337\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Ikegwu AC, Nweke FH, Anikwe CV (2024) Recent trends in computational intelligence for educational big data analysis. Iran J Comput Sci 7(1):103\u2013129.\n \n https://doi.org/10.1007/s42044-023-00158-5\n \n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Ikegwu AC, Nweke HF, Anikwe CV, Alo UR, Okonkwo OR (2022) Big data analytics for data-driven industry: a review of data sources, tools, challenges, solutions, and research directions. Clust Comput 25(5):3343\u20133387\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Ikegwu AC, Nweke HF, Mkpojiogu E, Anikwe CV, Igwe SA, Alo UR (2024) Recently emerging trends in big data analytic methods for modeling and combating climate change effects. Energy Inf 7(1):6\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Jan Z, Ahamed F, Mayer W, Patel N, Grossmann G, Stumptner M, Kuusk A (2023) Artificial intelligence for industry 4.0: systematic review of applications, challenges, and opportunities. Expert Syst Appl 216:119456\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Jang Y, Choi S, Jung H, Kim H (2022) Practical early prediction of students\u2019 performance using machine learning and eXplainable AI. Educ Inf Technol 27(9):12855\u201312889\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Jia J, Liang W, Liang Y (2023) A review of hybrid and ensemble in deep learning for natural language processing. arXiv preprint\n \n arXiv:2312.05589\n \n .\n
\n
\n
\n Kale A, Nguyen T, Harris FC Jr, Li C, Zhang J, Ma X (2023) Provenance documentation to enable explainable and trustworthy AI: a literature review. Data Intell 5(1):139\u2013162\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Kaminski ME (2021) The right to explanation, explained. In: Research Handbook on Information Law and Governance (pp 278\u2013299). Edward Elgar Publishing.\n
\n
\n
\n Kaushik R (2023) Explainability in machine learning: bridging the gap between model complexity and interpretability. Edu Journal of International Affairs and Research, ISSN\n \n :\n \n 2583\u20139993, 2(4), 57\u201363.\n
\n
\n
\n Kesari A, Sele D, Ash E, Bechtold S (2024) A Legal framework for explainable artificial intelligence. Center for Law & Economics Working Paper Series, 9.\n
\n
\n
\n Khosravi H, Shum SB, Chen G, Conati C (2022) Explainable artificial intelligence in education computers and education\u202f: artificial intelligence explainable artificial intelligence in education Dragan Ga \u02c7. May.\n \n https://doi.org/10.1016/j.caeai.2022.100074\n \n
\n
\n
\n Khrais LT (2020) Role of artificial intelligence in shaping consumer demand in E-commerce. Future Internet 12(12):226\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Kosasih E, Papadakis E, Baryannis G, Brintrup A (2023). Explainable artificial intelligence in supply chain management: a systematic review of neurosymbolic approaches. Int J Prod Res.\n
\n
\n
\n Krig S (2014) Computer vision metrics: survey, taxonomy, and analysis (p 508). Springer nature.\n
\n
\n
\n Krishnamoorthy CS, Rajeev S (2018) Artificial intelligence and expert systems for engineers. CRC Press\n
\n
\n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Kulesza T, Stumpf S, Herlocker J (2015) Designing explanatory interfaces. Proceedings of the 2015 international conference on intelligent user interfaces. Link\n
\n
\n
\n Kumar Y, Jain Y (2012) Research aspects of an expert system. Int J Comput Bus Res 1(11).\n \n https://api.semanticscholar.org/CorpusID:61842788\n \n
\n
\n
\n Ku\u017aniacki B, Almada M, Tyli\u0144ski K, G\u00f3rski \u0141, Winogradska B, Zeldenrust R (2022) Towards explainable artificial intelligence (XAI) in tax law: the need for a minimum legal standard. World Tax J 14:573\u2013616\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Kuznietsov A, Gyevnar B, Wang C, Peters S, Albrecht SV (2024). Explainable AI for safe and trustworthy autonomous driving: a systematic review. arXiv preprint\n \n arXiv:2402.10086\n \n .\n
\n
\n
\n Lakshmi V, Corbett J (2023). Using AI to improve sustainable agricultural practices: a literature review and research agenda.\n
\n
\n
\n Lawrence B (2019) The role of legal scholars in the development of artificial intelligence. Harv Law Rev Forum 132:157\u2013167\n
\n
\n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Lee A (2024) Knowledge representation and reasoning in AI: analyzing different approaches to knowledge representation and reasoning in artificial intelligence systems. J Artif Intell Res 4(1):14\u201329\n
\n
\n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Lee AVY, Koh E, Looi CK (2023) AI in education and learning analytics in Singapore: an overview of key projects and initiatives. Inf Technol Edu Learn 3(1):Inv-p001\n
\n
\n \n Google Scholar\n \n
\n
\n
\n Lim S (2021) Judicial decision-making and explainable artificial intelligence: a reckoning from first principles. Singap Acad Law J 33:280\u2013314\n
\n
\n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Linheiro ES, Shinde GR, Mahalle PN, Mirajkar R (2023) Explainable AI (XAI) for agriculture. Industry 4.0 convergence with AI, IoT, big data and cloud computing: fundamentals, challenges and applications, 161.\n
\n
\n
\n Lisboa PJ, Saralajew S, Vellido A, Fern\u00e1ndez-Domenech R, Villmann T (2023) The coming of age of interpretable and explainable machine learning models. Neurocomputing 535:25\u201339\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Li M, Wan Z, Zou T, Shen Z, Li M, Wang C, Xiao X (2024) Artificial intelligence enabled self-powered wireless sensing for smart industry. Chem Eng J 492:152417\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Liu J, Chen K, Lyu W (2024) Embracing artificial intelligence in the labour market: the case of statistics. Humanit Soc Sci Commun 11(1):1.\n \n https://doi.org/10.1057/s41599-024-03557-6\n \n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Li T, Li Y (2023) Artificial intelligence for reducing the carbon emissions of 5G networks in China. Nat Sustain 6(12):1522\u20131523\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Liu Q, Pinto JD, Paquette L (2024) Applications of explainable ai (xai) in education. In Trust and Inclusion in AI-Mediated Education: Where Human Learning Meets Learning Machines (pp. 93\u2013109). Cham: Springer Nature Switzerland.\n
\n
\n
\n Long S, Huang W, Wang J, Liu J, Gu Y, Wang Z (2024) A fixed-time consensus control with prescribed performance for multi-agent systems under full-state constraints. IEEE Transact Autom Sci Eng\n
\n
\n
\n Love PE, Fang W, Matthews J, Porter S, Luo H, Ding L (2023) Explainable artificial intelligence (XAI): precepts, models, and opportunities for research in construction. Adv Eng Inform 57:102024\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Lythe M, de Cos GM, Mingallon M, Lensen A, Galloway C, Knox D, Auvaa S, Kumarasinghe K (2023) Explainable AI\u2013building trust through understanding.\n
\n
\n
\n Mahbooba B, Timilsina M, Sahal R, Serrano M (2021) Explainable artificial intelligence (XAI) to enhance trust management in intrusion detection systems using decision tree model. Complexity 2021:1\u201311\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Majumdar S (2023) Fairness, explainability, privacy, and robustness for trustworthy algorithmic decision-making. In Big data analytics in chemoinformatics and bioinformatics (pp 61\u201395). Elsevier.\n
\n
\n
\n Mana AA, Allouhi A, Hamrani A, Rahman S, el Jamaoui I, Jayachandran K (2024) Sustainable AI-based production agriculture: exploring AI applications and implications in agricultural practices. Smart Agric Technol 17:100416\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Matcov A (2024) Explainable AI in Credit Risk Assessment for External Customers (Bachelor\u2019s thesis, University of Twente).\n
\n
\n
\n Mehta S, Rogers A, Gilbert TK (2023) Dynamic Documentation for AI Systems. arXiv preprint\n \n arXiv:2303.10854\n \n .\n
\n
\n
\n Minh D, Wang HX, Li YF, Nguyen TN (2022) Explainable artificial intelligence: a comprehensive review. Artificial Intelligence Review, 1\u201366.\n
\n
\n
\n Mittelstadt B, Russell C, Wachter S (2019) Explaining explanations in AI. In Proceedings of the conference on fairness, accountability, and transparency (pp 279\u2013288).\n
\n
\n
\n Mohammed AA, Ambak K, Mosa AM, Syamsunur D (2019) Expert system in engineering transportation: a review. J Eng Sci Technol 14(1):229\u2013252\n
\n
\n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Montavon G, Samek W, M\u00fcller KR (2018) Methods for interpreting and understanding deep neural networks. Digital Signal Process 73:1\u201315\n
\n
\n \n Article\n \n \n MathSciNet\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Nagahisarchoghaei M, Nur N, Cummins L, Nur N, Karimi MM, Nandanwar S, Rahimi S (2023) An empirical survey on explainable ai technologies: recent trends, use-cases, and categories from technical and application perspectives. Electronics 12(5):1092\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Nallakaruppan MK, Balusamy B, Shri ML, Malathi V, Bhattacharyya S (2024) An Explainable AI framework for credit evaluation and analysis. Appl Soft Comput 153:111307\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Nixon M, Aguado A (2019) Feature extraction and image processing for computer vision. Academic press.\n
\n
\n
\n Noguerol TM, Paulano-Godino F, Mart\u00edn-Valdivia MT, Menias CO, Luna A (2019) Strengths, weaknesses, opportunities, and threats analysis of artificial intelligence and machine learning applications in radiology. J Am Coll Radiol 16(9):1239\u20131247\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Noorbakhsh-Sabet N, Zand R, Zhang Y, Abedi V (2019) Artificial intelligence transforms the future of health care. Am J Med 132(7):795\u2013801\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Nott G (2017) Explainable artificial intelligence: Cracking open the black box of AI. Computer world, 4.\n
\n
\n
\n Nwakanma CI, Ahakonye LA, Njoku JN, Odirichukwu JC, Okolie SA, Uzondu C, Ndubuisi Nweke CC, Kim D (2023) Applied sciences explainable artificial intelligence ( XAI ) for intrusion detection and mitigation in intelligent connected vehicles a review. Appl Sci 13:1252\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Nwosu AC, Sturgeon B, McGlinchey T, Goodwin CD, Behera A, Mason S, Payne TR (2019) Robotic technology for palliative and supportive care: strengths, weaknesses, opportunities and threats. Palliat Med 33(8):1106\u20131113\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n O\u2019Mahony N, Campbell S, Carvalho A, Harapanahalli S, Hernandez GV, Krpalkova L, Riordan D, Walsh J (2020) Deep learning vs. traditional computer vision. In Advances in Computer Vision: Proceedings of the 2019 Computer Vision Conference (CVC), Volume 1 1 (pp. 128\u2013144). Springer International Publishing.\n
\n
\n
\n Panayides AS, Amini A, Filipovic ND, Sharma A, Tsaftaris SA, Young A, Pattichis CS (2020) AI in medical imaging informatics: current challenges and future directions. IEEE J Biomed Health Info 24(7):1837\u20131857\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Pawar U, O\u2019Shea D, Rea S, O\u2019Reilly R\u00a0(2020). Incorporating Explainable Artificial Intelligence (XAI) to aid the Understanding of Machine Learning in the Healthcare Domain. In\n \n Aics\n \n (pp 169\u2013180).\n
\n
\n
\n Plass M, Kargl M, Kiehl TR, Regitnig P, Gei\u00dfler C, Evans T, M\u00fcller H (2023) Explainability and causability in digital pathology. J Pathol: Clin Res 9(4):251\u2013260\n
\n
\n \n Google Scholar\n \n
\n
\n
\n Poch\u00e9 A, Hervier L, Bakkay MC (2023). Natural Example-Based Explainability: a Survey. In World Conference on explainable Artificial Intelligence (pp 24\u201347). Cham: Springer Nature Switzerland.\n
\n
\n
\n Posadas BB, Ogunyiola A, Niewolny K (2023) Socially responsible AI assurance in precision agriculture for farmers and policymakers. In AI Assurance (pp 473\u2013499). Academic Press.\n
\n
\n
\n Poursabzi-Sangdeh F, Goldstein DG, Hofman JM, Wortman Vaughan JW, Wallach H (2021) Manipulating and measuring model interpretability. In: Proceedings of the 2021 CHI conference on human factors in computing systems (pp. 1\u201352).\n
\n
\n
\n Pushkarna M, Zaldivar A, Kjartansson O (2022) Data cards: purposeful and transparent dataset documentation for responsible AI. In: Proceedings of the 2022 ACM conference on fairness, accountability, and transparency (pp 1776\u20131826).\n
\n
\n
\n Rachha A, Seyam M (2023) Explainable AI in education\u202f: current trends, challenges and opportunities. SoutheastCon 2023.\n \n https://doi.org/10.1109/SoutheastCon51012.2023.10115140\n \n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Ragno A, La Rosa B, Capobianco R (2022) Prototype-based interpretable graph neural networks. IEEE Transactions on Artificial Intelligence.\n
\n
\n
\n Rane N, Choudhary S, Rane J (2023) Explainable artificial intelligence (XAI) approaches for transparency and accountability in financial decision-making. Available at SSRN 4640316.\n
\n
\n
\n Rasheed MH, Khalid J, Ali A, Rasheed MS, Ali K (2024) Human resource analytics in the era of artificial intelligence: leveraging knowledge towards organizational success in Pakistan. J Chin Hum Resour Manag 15(3):3\u201320.\n \n https://doi.org/10.47297/wspchrmWSP2040-800501.20241503\n \n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Rong Y, Xu Z, Liu J, Liu H, Ding J, Liu X, Gao J (2022) Du-bus: a realtime bus waiting time estimation system based on multi-source data. IEEE Transact Intell Transp Syst 23(12):24524\u201324539\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Rajan K, Saffiotti A (2017) Towards a science of integrated AI and robotics. Artif Intell 247:1\u20139\n
\n
\n \n Article\n \n \n MathSciNet\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Reshamwala A, Mishra D, Pawar P (2013) Review on natural language processing. IRACST Eng Sci Technol: Int J (ESTIJ) 3(1):113\u2013116\n
\n
\n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Retzlaff CO, Angerschmid A, Saranti A, Schneeberger D, Roettger R, Mueller H, Holzinger A (2024) Post-hoc vs ante-hoc explanations: xAI design guidelines for data scientists. Cogn Syst Res 86:101243\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Ryo M (2022) Explainable artificial intelligence and interpretable machine learning for agricultural data analysis. Artif Intell Agric 6:257\u2013265\n
\n
\n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Sabrina F, Sohail S, Farid F, Jahan S, Ahamed F, Gordon S (2022) An interpretable artificial intelligence-based smart agriculture system. Comput Mater Continua 72:3777\u20133797\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Sadeghi K, Ojha D, Kaur P, Mahto RV, Dhir A (2024) Explainable artificial intelligence and agile decision-making in supply chain cyber resilience. Decis Support Syst 180:114194\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Saeed W, Omlin C (2023) Explainable AI (XAI): a systematic meta-survey of current challenges and future opportunities. Knowl-Based Syst 263:110273\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Samek W, Wiegand T, M\u00fcller KR (2017) Explainable artificial intelligence: understanding, visualizing, and interpreting deep learning models. arXiv preprint\n \n arXiv:1708.08296\n \n .\n
\n
\n
\n Sarker IH (2022) AI-based modeling: techniques, applications and research issues towards automation, intelligent and smart systems. SN Computer Science 3(2):158\n
\n
\n \n Article\n \n \n MathSciNet\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Schoenherr JR, Abbas R, Michael K, Rivas P, Anderson TD (2023) Designing AI using a human-centered approach: Explainability and accuracy toward trustworthiness. IEEE Transact Technol Soc 4(1):9\u201323\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Schoonderwoerd TA, Jorritsma W, Neerincx MA, Van Den Bosch K (2021) Human-centered XAI: developing design patterns for explanations of clinical decision support systems. Int Lodg Ind: Orig Last Front 154:279\u2013313\n
\n
\n \n Google Scholar\n \n
\n
\n
\n Stepin I, Alonso JM, Catala A, Pereira-Fari\u00f1a M (2021) A survey of contrastive and counterfactual explanation generation methods for explainable artificial intelligence. IEEE Access 9:11974\u201312001\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Stiglic G, Kocbek P, Fijacko N, Zitnik M, Verbert K, Cilar L (2020) Interpretability of machine learning-based prediction models in healthcare. Wiley Interdiscip Rev: Data Min Knowl Discov 10(5):e1379\n
\n
\n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Sunitha B, Kiran BK (2023) A systematic review of recommendation systems: applications and challenges. Int J Intell Syst Appl Eng 11(2):826\u2013846\n
\n
\n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Sutherland A, Wright D (2020) The Governance of Artificial Intelligence in the Legal Sector. AI & Society, 35(1), 119\u2013133. Link\n
\n
\n
\n Taghikhah F, Voinov A, Filatova T, Polhill JG (2022) Machine-assisted agent-based modeling: opening the black box. J Comput Sci 64:101854\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Talaat FM, Aljadani A, Badawy M, Elhosseini M (2024) Toward interpretable credit scoring: integrating explainable artificial intelligence with deep learning for credit card default prediction. Neural Comput Appl 36(9):4847\u20134865\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Torfi A, Shirvani RA, Keneshloo Y, Tavaf N, Fox EA (2020) Natural language processing advancements by deep learning: A survey. arXiv preprint\n \n arXiv:2003.01200\n \n .\n
\n
\n
\n Tucci V, Saary J, Doyle TE\u00a0(2022) Factors influencing trust in medical artificial intelligence for healthcare professionals: A narrative review. J Med Artif Intell, 5.\n
\n
\n
\n Valavanis KP, Saridis GN (2012) Intelligent robotic systems: theory, design, and applications (vol 182). Springer Science & Business Media.\n
\n
\n
\n Vale D, El-Sharif A, Ali M (2022) Explainable artificial intelligence (XAI) post-hoc explainability methods: risks and limitations in non-discrimination law. AI and Ethics 2(4):815\u2013826\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Vilone G, Rizzo L, Longo L (2020) A comparative analysis of rule-based, model-agnostic methods for explainable artificial intelligence.\n
\n
\n
\n Vinod B (2022) Artificial intelligence and emerging technologies in hospitality. revenue management in the lodging industry: origins to the last frontier, 279\u2013313.\n
\n
\n
\n Vollert S, Atzmueller M, Theissler A (2021) Interpretable Machine Learning: A brief survey from the predictive maintenance perspective. In 2021 26th IEEE international conference on emerging technologies and factory automation (ETFA) (pp 01\u201308). IEEE.\n
\n
\n
\n Voj\u00ed\u0159 S, Kliegr T (2020) Editable machine learning models? A rule-based framework for user studies of explainability. Adv Data Anal Classif 14(4):785\u2013799\n
\n
\n \n Article\n \n \n MathSciNet\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Voosen P (2017). The AI detectives.\n
\n
\n
\n Voulodimos A, Doulamis N, Doulamis A, Protopapadakis E (2018). Deep learning for computer vision: A brief review. Computational intelligence and neuroscience, 2018.\n
\n
\n
\n Vyas B (2023) Explainable AI: assessing methods to make AI systems more transparent and interpretable. Int J New Med Stud: Int Peer Rev Sch Index J 10(1):236\u2013242\n
\n
\n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Walker CM, Agarwal V, Lin L, Hall AC, Hill RA, Laurids R, Mortenson TJ, Lybeck NJ (2023) Explainable Artificial Intelligence Technology for Predictive Maintenance (No. INL/RPT-23-74159-Rev000). Idaho National Laboratory (INL), Idaho Falls, ID (United States).\n
\n
\n
\n Wang E, Yang Y, Wu J, Liu W, Wang X (2018) An efficient prediction-based user recruitment for mobile crowdsensing. EEE Transact Mob Comput 17(1):16\u201328\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Wang P, Wei Z, Qi H, Wan S, Xiao Y, Sun G, Zhang Q (2023) Mitigating poor data quality impact with federated unlearning for human-centric metaverse. IEEE J Sel Areas Commun\n
\n
\n
\n Wang XS, Ryoo JHJ, Bendle N, Kopalle PK (2021) The role of machine learning analytics and metrics in retailing research. J Retail 97(4):658\u2013675\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Wang Y, Chandrasekaran J, Haberkorn F, Dong Y, Gopinath M, Batarseh FA (2022) Deepfarm: AI-driven management of farm production using explainable causality. In 2022 IEEE 29th annual software technology conference (STC) (pp 27\u201336). IEEE.\n
\n
\n
\n Weber L, Lapuschkin S, Binder A, Samek W (2023) Beyond explaining: Opportunities and challenges of XAI-based model improvement. Inf Fus 92:154\u2013176\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Weber P, Carl KV, Hinz O (2023) Applications of explainable artificial intelligence in finance\u2014a systematic review of finance, information systems, and computer science literature. Management Review Quarterly, 1\u201341.\n
\n
\n
\n Wu L, Long Y, Gao C, Wang Z, Zhang Y (2023) MFIR: multimodal fusion and inconsistency reasoning for explainable fake news detection. Inf Fus 100:101944\n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Zhang S, Li T, Hui S, Li G, Liang Y, Yu L, Jin D, Li Y (2023) Deep Transfer Learning for City-scale Cellular Traffic Generation through Urban Knowledge Graph. Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 4842\u20134851.\n \n https://doi.org/10.1145/3580305.3599801\n \n
\n
\n
\n Wu K, Chi K (2023) Enhanced e-commerce customer engagement: A comprehensive three-tiered recommendation system. Journal of Knowledge Learning and Science Technology ISSN: 2959\u20136386 (online), 2(3), 348\u2013359.\n
\n
\n
\n Yang Y, Zhuang Y, Pan Y (2021) Multiple knowledge representation for big data artificial intelligence: framework, applications, and case studies. Front Inf Technol Electron Eng 22(12):1551\u20131558\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Zacharias J, von Zahn M, Chen J, Hinz O (2022) Designing a feature selection method based on explainable artificial intelligence. Electron Mark 32(4):2159\u20132184\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Zhang Y, Weng Y, Lund J (2022) Applications of explainable artificial intelligence in diagnosis and surgery. Diagnostics 12(2):237\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n
\n
\n Zhu C (2024) Computational intelligence-based classification system for the diagnosis of memory impairment in psychoactive substance users. J Cloud Comput.\n \n https://doi.org/10.1186/s13677-024-00675-z\n \n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Zou X, Yuan J, Shilane P, Xia W, Zhang H, Wang X (2022) From hyper-dimensional structures to linear structures: maintaining deduplicated data\u2019s locality. ACM Transact Storage.\n \n https://doi.org/10.1145/3507921\n \n
\n
\n \n Article\n \n \n Google Scholar\n \n
\n
\n
\n Zhuang YT, Wu F, Chen C, Pan YH (2017) Challenges and opportunities: from big data to knowledge in AI 2.0. Front Inf Technol Electron Eng 18:3\u201314\n
\n
\n \n Article\n \n \n MATH\n \n \n Google Scholar\n \n
\n

\n \n Download references\n \n

\n Funding\n

\n There is no external funding received by the authors.\n

\n Author information\n

\n Authors and Affiliations\n

\n
\n Centre for Atmospheric Research, National Space Research and Development Agency, Anyigba, Nigeria\n
\n
\n Daniel Enemona Mathew\n
\n
\n
\n Computer Engineering Department, Faculty of Engineering, Veritas University, Abuja, Nigeria\n
\n
\n Deborah Uzoamaka Ebem\n
\n
\n
\n Software Engineering Department, Faculty of Natural and Applied Sciences, Veritas University, Abuja, Nigeria\n
\n
\n Anayo Chukwu Ikegwu\n
\n
\n
\n Computer Science Department, Faculty of Physical Sciences, University of Nigeria, Nsukka (UNN), Nigeria\n
\n
\n Daniel Enemona Mathew,\u00a0Deborah Uzoamaka Ebem,\u00a0Pamela Eberechukwu Ukeoma\u00a0&\u00a0Ngozi Fidelia Dibiaezue\n
\n

\n Authors\n

\n Daniel Enemona Mathew\n
\n \n View author publications\n \n
\n Search author on:\n \n \n PubMed\n \n \n Google Scholar\n \n \n
\n
\n
\n Deborah Uzoamaka Ebem\n
\n \n View author publications\n \n
\n Search author on:\n \n \n PubMed\n \n \n Google Scholar\n \n \n
\n
\n
\n Anayo Chukwu Ikegwu\n
\n \n View author publications\n \n
\n Search author on:\n \n \n PubMed\n \n \n Google Scholar\n \n \n
\n
\n
\n Pamela Eberechukwu Ukeoma\n
\n \n View author publications\n \n
\n Search author on:\n \n \n PubMed\n \n \n Google Scholar\n \n \n
\n
\n
\n Ngozi Fidelia Dibiaezue\n
\n \n View author publications\n \n
\n Search author on:\n \n \n PubMed\n \n \n Google Scholar\n \n \n
\n
\n

\n Contributions\n

\n DEM coined the title, surfed the internet for materials, and participated in manuscript preparation especially Sect. 2, 3 and 4. ACI drafted out the structures, wrote \u201cThe introduction, and conclusion sections, and participated in the reviewing of the entire manuscript sections. DUE participated in the drafting, grammatical check, correction, and editing. PEU screened the materials and participated in writing the \u201cExplanatory Depth of Explainable AI\u201d section. In addition, NFD participated in the manuscript draft, proofreading, and reviewing of the entire manuscript.\n

\n Corresponding authors\n

\n Correspondence to\n \n Daniel Enemona Mathew\n \n or\n \n Anayo Chukwu Ikegwu\n \n .\n

\n Ethics declarations\n

\n Conflict of interest\n

\n The authors declared that there is no conflict of interest in this article.\n

\n Ethical Approval and Consent to Participate\n

\n Not applicable.\n

\n Consent for Publication\n

\n Not applicable.\n

\n Additional information\n

\n Publisher's Note\n

\n Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.\n

\n Rights and permissions\n

\n \n Open Access\n \n This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article\u2019s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article\u2019s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit\n \n http://creativecommons.org/licenses/by-nc-nd/4.0/\n \n .\n

\n \n Reprints and permissions\n \n

\n About this article\n

\n Cite this article\n

\n Mathew, D.E., Ebem, D.U., Ikegwu, A.C.\n \n et al.\n \n Recent Emerging Techniques in Explainable Artificial Intelligence to Enhance the Interpretable and Understanding of AI Models for Human.\n \n Neural Process Lett\n \n \n 57\n \n , 16 (2025). https://doi.org/10.1007/s11063-025-11732-2\n

\n \n Download citation\n \n

\n
\n Accepted\n :\n \n 21 January 2025\n \n
\n
\n
\n Published\n :\n \n 07 February 2025\n \n
\n
\n
\n \n DOI\n \n :\n https://doi.org/10.1007/s11063-025-11732-2\n
\n

\n Share this article\n

\n Anyone you share the following link with will be able to read this content:\n

\n \n

\n Sorry, a shareable link is not currently available for this article.\n

\n \n

\n Provided by the Springer Nature SharedIt content-sharing initiative\n

\n Keywords\n

\n \n\n", "best_field": "cleaned_text_custom_parser", "enhanced": true, "date_published": null, "image": null, "num_tokens_oai_tokenizer": null, "num_tokens_gemini_tokenizer": null}, "favicon": "https://link.springer.com/oscar-static/img/favicons/darwin/apple-touch-icon-92e819bf8a.png", "state": "candidate", "enhanced": true, "answer": null, "valid_answer": false, "source_citation_index": 7, "user_web_page_id": null, "status": "ready", "reader": false, "rank": 2, "reason": "The source covers challenges and emerging techniques in explainable AI, which indirectly relates to explainability standards."}, {"id": 605125, "type": "GenericWebPage", "mime_type": "text/html", "url": "https://quality.arc42.org/qualities/explainability", "slug": "arc42", "title": "Explainability | arc42 Quality Model", "snippet": "Explainability is a quality sometimes required in the context of Artificial Intelligence (AI). The European\u2019s proposed AI Act will probably contain obligations for some AI systems to provide explanations about decisions that impact user rights. Definitions A system S is explainable with respect ...", "image": null, "date_published": null, "media": [], "products": [], "source": {"id": 605125, "website_id": 133721, "url": "https://quality.arc42.org/qualities/explainability", "title": "Explainability | arc42 Quality Model", "snippet": "Explainability is a quality sometimes required in the context of Artificial Intelligence (AI). The European\u2019s proposed AI Act will probably contain obligations for some AI systems to provide explanations about decisions that impact user rights. Definitions A system S is explainable with respect ...", "cleaned_text": null, "cleaned_text_custom_parser": null, "cleaned_text_custom_parser_2": null, "best_field": null, "enhanced": false, "date_published": null, "image": null, "num_tokens_oai_tokenizer": null, "num_tokens_gemini_tokenizer": null}, "favicon": null, "state": "candidate", "enhanced": false, "answer": null, "valid_answer": false, "source_citation_index": 8, "user_web_page_id": null, "status": "not_processed", "reader": false, "rank": 2, "reason": "The document discusses the need for transparency and explainability in AI systems and mentions existing guidelines, though not specifically standards."}]