Should arXiv reject an AI-discovered cancer breakthrough that passes clinical trials?

IanArawjo · x · 2026-09-14

Ian Arawjo sparked a debate on X about AI-assisted research: if someone solved a type of cancer through clever AI use—validated repeatedly in empirical tests and clinical trials, even if not fully understood—should arXiv reject the paper? He argues that would be absurd, and pushes back on the idea that "incomplete understanding" means no human effort: even cutting-edge math relies on heavy human oversight and hunches.

Original post →

More from AGI Musings

AGI Musings channel →