Reproduce “Beyond Explainable AI (XAI): An Overdue Paradigm Shift and Post-XAI Research Directions” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Saleh Afroogh, Syed Ishtiaque Ahmed, Petra Ahrweiler, David Alvarez-Melis, Mansur Maturidi Arief, Emilia Barakova, Falco J. Bargagli-Stoffi, Erdem Biyik and 41 more
The authors argue that post-hoc explanations of complex AI models are not proof that recommendations are reliable. They propose expert collaboration and domain-specific verification, and continued study of model internals. In a cited study of 457 clinicians diagnosing acute respiratory failure, accuracy was 9.1 percentage points lower in a biased-AI-with-explanations setting; the paper does not test whether its proposals improve outcomes.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.