LLMs Exacerbate p-hacking in Scientific Research

krisgligoric · x · 2026-07-06

The author notes that while p-hacking is a long-standing issue in the scientific community, LLMs (when used as annotation tools or LLM-as-a-judge) make it even easier to achieve desired results simply by tweaking parameters. The author proposes a mitigation method and verifies its effectiveness through a pre-registered study.

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

Original post →

More from Research

Research channel →