Year: 2026
Type: preprint
Abstract: This study provides a cross-disciplinary examination of Explainable Artificial Intelligence (XAI) approaches-focusing on deep neural networks (DNNs) and large language models (LLMs)-and identifies empirical and conceptual limitations in current XAI. We discuss critical symptoms that stem from deeper root causes (i.e., two paradoxes, two conceptual confusions, and five false assumptions). These fundamental problems within the current XAI research field reveal three insights: experimentally, XAI exhibits significant flaws; conceptually, it is paradoxical; and pragmatically, further attempts to reform the paradoxical XAI might exacerbate its confusion-demanding fundamental shifts and new research directions. To move beyond XAI's limitations, we propose a four-pronged synthesized paradigm shift toward reliable and certified AI development. These four components include: verification-focused Interactive AI (IAI) to establish scientific community protocols for certifying AI system performance rather than attempting post-hoc explanations, AI Epistemology for rigorous scientific foundations, User-Sensible AI to create context-aware systems tailored to specific user communities, and Model-Centered Interpretability for faithful technical analysis-together offering comprehensive post-XAI research directions.
Source: arXiv (Cornell University)
Authors: Afroogh, Saleh, Ahmed, Syed Ishtiaque, Ahrweiler, Petra, Alvarez-Melis, David, Arief, Mansur Maturidi, Barakova, Emilia, Bargagli-Stoffi, Falco J., Biyik, Erdem, Chen, Hanjie, Chen, Xiang 'Anthony', Clements, Robert Alan, Crockett, Keeley, Dhurandhar, Amit, Dogan, Fethiye Irmak, Dollinger, Mollie, Eslami, Motahhare, Faisal, Aldo A, Farahi, Arya, Pradier, Melanie F., Gabriel, Saadia, Garcia-Olano, Diego, Ghassemi, Marzyeh, Ghosh, Shaona, Gunes, Hatice, Hajiramezanali, Ehsan, Haufe, Stefan, Huang, Biwei, Hwang, Angel, Islam, Md Tauhidul, Jiao, Junfeng, Karimi, Amir-Hossein, Kazeminasab, Saber, Kuzminykh, Anastasia, La Cava, William, Lim, Brian Y., Liu, Xiaofeng, Mofrad, Mohammad R. K., Parrish, Alicia, Perez-Ortiz, Maria, Raj, Shriti, Swayamdipta, Swabha, Talebi, Salmonn, Varshney, Kush R., Vorvoreanu, Mihaela, Weng, Lily, Xiang, Alice, Xu, Yiming, Zhao, Ding, Zhao, Jieyu
Institutions: The University of Texas at Austin, University of Applied Sciences Mainz, Johannes Gutenberg University Mainz, Harvard University, Stanford University, Eindhoven University of Technology, University of California, Los Angeles, Rice University, University of San Diego, University of California San Diego, University of San Francisco, Manchester Metropolitan University, Curtin University, Imperial College London, Carnegie Mellon University, University of Cambridge, Bridge University, Microsoft (United States), Microsoft Research (United Kingdom), Nvidia (United Kingdom), Massachusetts Institute of Technology, Technische Universität Berlin, University of Toronto, Lawrence Berkeley National Laboratory, Yale University, University of California, Berkeley, SOAS University of London, University College London
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Funders: National Science Foundation, Gordon and Betty Moore Foundation, Harvard University, UK Research and Innovation, National Research Foundation, National Research Foundation Singapore, Info-communications Media Development Authority, Natural Sciences and Engineering Research Council of Canada, Economic and Social Research Council
Awards: G2066398493, NRT-AI: Convergent, Responsible, and Ethical Artificial Intelligence Training Experience for Roboticists, CHANSE-WB-941 Measuring children’s wellbeing and mental health with social robots (MICRO), CAREER: Ethical Machine Learning in Health: Robustness in Data, Learning and Deployment, G6671297155, G7622352226, AI Institute for Societal Decision Making (AI-SDM), NSF-Simons AI Institute for Cosmic Origins
OpenAlex ID: W7133228188