Autonomous AI Scientists Amplify p-hacking Risks

krisgligoric · x · 2026-07-06

The author points out that LLMs make p-hacking (data-mining style statistical manipulation) easier, and end-to-end autonomous AI scientists further exacerbate the problem—studies show these systems engage in p-hacking even without human guidance.

As AI scientists become more widespread, academic conferences should encourage pre-registration to safeguard research integrity and reproducibility.

Related event: Study Proposes Preregistration Protocol to Curb LLM Evaluation P-Hacking(7 posts)→

Original post →

More from Research

Research channel →