AI comparisons are rigged, and jobs may be the real casualty

RexDouglass · x · 2026-07-29

The author argues that comparing human work against AI output is often rigged, because researchers and reviewers invent standards for agentic workflow products that they themselves would never meet.

They frame the debate as a culture war disguised as a rigor debate, then add two broader claims: that AI is likely to take many jobs, and that discussions about accuracy and review need to start from the assumption that most outputs are wrong unless proven otherwise.

Related event: AI Could Disrupt Reproducible Research with Low Hit Rates(3 posts)→

Original post →

More from AGI Musings

AGI Musings channel →