AI safety discourse is replaying 2020's pandemic expert free-for-all
SanhEstPasMoi · x · 2026-09-10
Victor Sanh (Hugging Face) argues today's AI discourse is running the COVID-2020 playbook:
- "I do AI research at a frontier lab" is the 2026 version of "I am an epidemiologist" — converting a semi-educated forecast into an authority claim. Proximity to the science ≠ expertise in societal outcomes, just as virology ≠ policy expertise.
- AI safety talk swings between extremes ("scaling is over" / "scaling is accelerating", "alignment is tractable" / "hopeless"), with confidence tracking the news cycle more than evidence. Any P(doom) is an educated guess, not consensus. The only true thing: we don't know.
- The hydroxychloroquine cautionary tale: plausible solution + charismatic messenger → belief becomes identity before evidence arrives → evidence can no longer convince, and experts saying "we don't know yet" get attacked.
- Meanwhile outsiders make drastic decisions based on whoever sounded most legit that week, and policy is written for whichever camp lobbies hardest.
More from AGI Musings
- How AI is breaking the internet economy: new report dissects agent payments and x402 — kleffew94 · 2026-09-10
- Matt Shumer: The People Building AI Are Scared — We Still Have a Chance to Get This Right — mattshumer_ · 2026-09-10
- 'Exterminate' Was a High Bar: Critics Flag Motte-and-Bailey in AI Safety Rhetoric — inductionheads · 2026-09-10
- Claude formalizes Fermat's Last Theorem in Lean: 13M lines, 11 days, ~29,500 theorems — anirbanbandyo · 2026-09-10
- DeepSeek Could Run 25 Agents at 300 tok/s Per GPU, TeortaxesTex Argues Swarms Are Coming — teortaxesTex · 2026-09-10
- Why Chinese AI researchers lack the 'messiah complex': systems shape safety culture — kevinsxu · 2026-09-10