Explainable AI Across Domains: Techniques, Domain-Specific Applications, and Future Directions
Xinyuan Song, Tianyang Wang, Ming Liu, Yunze Wang, Benji Peng, Silin Chen, Qian Niu, Junyu Liu, Keyu Chen, Ming Li, Pohsun Feng, Ziqian Bi, Yichao Zhang, Cheng Fei, Lawrence K.Q. Yan · 2024
Explainability in artificial intelligence (AI) has become crucial for ensuring transparency, trust, and usability across diverse application domains, such as healthcare, finance, and autonomous systems. This comprehensive review analyzes the state of research on explainability techniques, categorizing approaches into model-agnostic, model-specific, and hybrid methods. Key techniques, such as SHAP, LIME, and rule-based explanations, are discussed alongside their respective strengths and limitations. The review also delves into domain-specific applications, highlighting unique interpretability requirements in sectors like medical diagnostics, credit scoring, and autonomous decision-making. We further explore the evaluation metrics and benchmarks essential for assessing the quality and effectiveness of explainable AI, addressing challenges such as computational complexity, user-centered design, and ethical considerations. By identifying gaps in current methodologies, this review proposes future research directions aimed at developing adaptable, cross-domain explainability frameworks, enhancing robustness against adversarial manipulations, and promoting ethically aligned AI.