RL veteran Szepesvári models AI risk by swapping 'AI' for 'a human'
CsabaSzepesvari · x · 2026-09-14
Csaba Szepesvári discusses how to think about AI risk by substituting "AI" with "a human" or "a powerful organization," since human intelligence is the only pluripotent technology we have centuries of guardrail experience with (education, institutions, laws). He fails to see how a single very intelligent person poses existential risk; an army of cheap amoral ones raises the odds, but the risks he sees are mostly non-existential — and in that scenario a human is still in charge.
More from AGI Musings
- Commentary: frontier AI leaders don't trust each other, so they ask government to chaperone — bilawalsidhu · 2026-09-14
- Cognitive Scientist's Essay: Rethinking Human-AI Relations After Cyberattacks and Navier-Stokes — DistractedDendrite · 2026-09-14
- TCS researcher reflects on problem-solving vs theory-building in the AI era — jasondeanlee · 2026-09-14
- Pedro Domingos: The future of SaaS is SAIS — 'Software as an AI Service' — pmddomingos · 2026-09-14
- Pedro Domingos: Limited human-machine communication bandwidth is AI's biggest obstacle — pmddomingos · 2026-09-14
- Domingos: 'AI is eating software faster' is as weak a claim as 'software is eating hardware faster' — pmddomingos · 2026-09-14