The Emperor's New Algorithms: Why 99.3% of "Explainable" AI Has Never Met a HumanMIT Study Reveals Only 0.7% of Explainable AI Papers Actually Test on HumansJan 25, 2026·8 min read·5