Debate: Is AI Eval/Safety Work Really Only for a Tiny Talent Pool?

JacquesThibs · x · 2026-09-18

Sentdex pushes back on the claim that only "very few" people can work on evals/safety/alignment — arguing the perceived scarcity comes from counting only the 60-80 brilliant folks aligned with EA-style danger assessments. JacquesThibs responds that the real bottleneck is institutional: outsiders didn't build eval capacity because they weren't focused on AI risk, and shares his contract work building an app for CG that successfully surfaced non-EA researchers for funding.

Related event: Sentdex Disputes Claim That Only a Few Can Do AI Eval Work(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →