How to read shaky AI-job-impact studies: Bayesian updating beats gold standard

alexolegimas · x · 2026-09-21

A methodology debate over a paper linking AI exposure to weaker early-career employment: alexolegimas argues no such paper has gold-standard methodology — read them like a Bayesian, where six papers from different angles showing the same thing justify updating beliefs, since good causal data won't exist until the results are obsolete. Economist DavidSimon is skeptical, noting messy pre-trends in employment-to-population ratios and effect timing that doesn't line up with when AI should matter. danielrock adds that publication norms force papers to overclaim causality; the honest framing is 'AI-exposed early-career workers are having a tougher time, causality unclear.'

Related event: Debate Erupts Over Methodology of Studies on AI's Impact on Entry-Level Jobs(5 posts)→

Original post →

More from AGI Musings

AGI Musings channel →