AI Now researcher calls AI extinction claims unscientific: neither falsifiable nor verifiable
AINowInstitute · x · 2026-09-16
A Guardian explainer examines six recent AI-doom claims. AI Now's Heidy Khlaaf says existential-threat claims are "neither falsifiable nor verifiable, making them unscientific." Gary Marcus and colleagues push back on Dario Amodei's botnet-takeover warning, citing the UK AI Security Institute's finding that frontier models can only compromise weakly defended systems, with core internet infrastructure held by Cloudflare, Google and AWS.
Related event: Anthropic CEO's AI Botnet Warning Draws Expert Skepticism(4 posts)→
More from AGI Musings
- WSJ subhed of the day: talk of heading off AI's role in destroying humanity hurt some stocks, helped others — birchlse · 2026-09-16
- Philosopher Carissa Véliz on AI doom predictions: the future doesn't get predicted — it gets produced — CarissaVeliz · 2026-09-16
- DeepMind co-founder Shane Legg launches the DeepMind Institute for interdisciplinary AGI debate — ShaneLegg · 2026-09-16
- DeepMind Institute: frontier labs to share more on governing the AGI transition, says Allan Dafoe — ShaneLegg · 2026-09-16
- Berkeley workshop readout: Why public participation in AI feedback is costly — beenwrekt · 2026-09-16
- Gates: AI could make Indian healthcare affordable, but patient data is the hard part — TechNadu · 2026-09-16