AI Systems Outperform Human Economists in Causal Analysis Study

AI Systems Outperform Human Economists in Causal Analysis Study
A new research paper reveals that agentic AI systems are matching and exceeding human economists in causal inference tasks. The study compared AI models including GPT-5.4, GPT-5.3-Codex, and Claude Opus 4.6 against human researchers performing identical economic analysis work. Results show AI systems produced similar median estimates to humans but with notably tighter distribution ranges, suggesting fewer hallucinations and more consistent outputs. In a peer review tournament where AI models evaluated submissions from both humans and other AI systems, GPT-5.4 ranked first, followed by GPT-5.3-Codex and Claude Opus 4.6, with human researchers placing fourth.