AI Safety Researcher: Claiming 'Maybe We Don't Even Have 4 Months' Backfires
S_OhEigeartaigh · x · 2026-10-05
Responding to Rob Bensinger's 'Do we even have 120 days?' post, Séán Ó hÉigeartaigh argues that while the world underestimates the urgency of the AI situation, claiming we might not even have 4 months has big downsides: it's likely wrong, and it effectively rules out most interventions and processes with the best chance of actually working.
More from AGI Musings
- Peter Norvig gives a talk on AI progress — mariofilhoml · 2026-10-05
- Can physics get a proof checker? Lanyon AI CEO on why AI hasn't cracked physics yet — TOEwithCurt · 2026-10-05
- Interpretability researcher amplifies sarcastic list of everything 'EAs got right' — NeelNanda5 · 2026-10-05
- Acemoglu questions AI's health promises: even a cancer cure won't hit the clinic fast — avicgoldfarb · 2026-10-05
- Cambridge's Leverhulme CFI launches interdisciplinary PhD in AI and Society for October 2027 — dioscuri · 2026-10-05
- Phil Venables: AI Is Turning Pyramid Organizations Into Diamonds and Breaking Talent Pipelines — philvenables · 2026-10-05