Economists call for an epistemological shift in studying AI's economic impact

soumitrashukla9 · x · 2026-09-22

Economists are debating methodology for studying AI's economic effects. Alex Olegimas argues that while clean causal identification defines top-journal economics, such studies take years or decades — and we need signals now, even if imprecise. Soumitra Shukla agrees, adding that AI capabilities are step-changing between quarters (2026Q1 vs 2026Q3), making fixation on precise identification less useful. He echoes Daniel Rock on mismatched research norms and Anna Stansbury's call to give "collages of evidence" greater weight.

Related event: Economists Debate How to Read Imperfect Evidence on AI's Impact on Entry-Level Jobs(13 posts)→

Original post →

More from AGI Musings

AGI Musings channel →