As artificial intelligence (AI) systems become increasingly integral to decision-making across industries, particularly in high-stakes domains such as healthcare, finance, and autonomous vehicles, there is a pressing need for explainable models that enhance transparency, interpretability, and accountability. The field of Explainable Artificial Intelligence (xAI) has emerged to address the opacity of complex “black-box” models, yet existing literature lacks a structured overview that organizes xAI techniques according to specific machine learning paradigms. This survey addresses the gap in existing literature by investigating how xAI techniques can be systematically organized and applied across different machine learning paradigms, including Supervised Learning, Unsupervised Learning, Reinforcement Learning, Computer Vision, and Generative AI models. The research problem encompasses understanding the relationships between xAI methods and their applicability to various AI model types, as well as identifying key challenges and limitations in current xAI implementations. This paper aims to provide a comprehensive survey that offers a structured overview of xAI techniques tailored to specific machine learning paradigms, serving as both a primer and a reference for researchers, AI developers, and industry professionals seeking to implement explainability in their AI-driven systems. This survey synthesizes and organizes the extensive body of existing xAI literature according to machine learning paradigms, examining both ante-hoc and post-hoc explainability approaches while exploring their applications across diverse industries including healthcare, retail, finance, and manufacturing. Through this comprehensive review, several critical patterns emerge from the literature: a significant gap exists regarding generalizable metrics to compare and evaluate xAI techniques, making it difficult to assess which methods outperform others; fundamental trade-offs persist between model interpretability and accuracy, with transparent models still being preferred when explainability heavily outweighs accuracy requirements; the field lacks standardized evaluation frameworks despite recent attempts to develop comprehensive metrics; and xAI adoption faces challenges related to generating explanations digestible for non-technical stakeholders. The synthesis reveals that xAI techniques can be effectively organized by machine learning paradigm, providing practitioners with a structured approach to selecting appropriate explanation methods based on their specific model types. This comprehensive survey provides a detailed taxonomy of xAI methods organized by machine learning paradigm, addressing a gap in current survey literature. By linking xAI techniques to specific ML paradigms and exploring their applications across industries, this work supports the enhancement of trustworthiness and effectiveness of AI systems. The identification of key challenges, particularly the lack of standardized evaluation metrics and the interpretability-accuracy trade-off, highlights critical areas for future research and development in the field of explainable AI. • Surveyed recently published literature on explainable artificial intelligence. • Emphasizes the importance of integrating domain knowledge into the xAI framework. • Showcases real-world case studies and industrial applications. • Provides a foundational survey for future research work in the field.
Wilkinson et al. (Sun,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: