AI Audits Challenge Saez–Zucman

Two independent AI auditors rechecked Saez and Zucman’s papers (and replication files) across dozens of studies and report hundreds of errors, including dozens the auditors judge large enough to materially change headline numbers. If confirmed, these findings would force a reassessment of influential inequality estimates and the policy prescriptions that rely on them. — Shows AI can systematically re‑evaluate high‑stakes academic claims, shifting authority in public policy debates and potentially overturning widely cited empirical results.