AI researcher Jacy Anthis lectures at UC Berkeley on AI Safety and Digital Minds
jacyanthis · x · 2026-10-08
AI ethics researcher Jacy Anthis shared slides from his UC Berkeley lecture "AI Safety and Digital Minds," framing humanity at the vertex of the lightcone.
Key points:
- AI welfare debates will center on solidarity, not charity: moral concern will hinge on whether people see AI agents as ingroup members capable of mutual benefit and harm. Agents need no body or qualia—social capabilities like moral judgment, emotional expression, and cooperation matter most.
- Public opinion: per the AIMS survey, 17% of US adults already think some current AIs are sentient and 40% are unsure; most people outside Twitter/X don't dismiss digital consciousness.
- Growing tension: people feel increasingly threatened by digital minds and decreasingly view AI as a moral patient—expect polarization along this axis.
- Research gap: we lack foundational social theory (like supply and demand) for varying forms of human–AI coexistence.
- Against P(doom): he finds it unhelpful; futures include states worse than extinction, different extinctions, and many ways to flourish. He proposes E(boom)—the expected moral value of the current boom.
More from AGI Musings
- P=NP classroom poll flip: from one believer in 2022 to a third of the class in 2025 — aran_nayebi · 2026-10-08
- "Cope all you want": the case that LLM progress compounds even if papers go unread — basedjensen · 2026-10-08
- Economist John J. Horton: AI will usher in a golden age for human research — soumitrashukla9 · 2026-10-08
- Economist John Horton: AI models would have saved teenage me months of being stuck — soumitrashukla9 · 2026-10-08
- Will the leverage live in superintelligent general models or in specialized harnesses? — zeeg · 2026-10-08
- gfodor: skills are largely teachable, so AI will eat both ends of the labor distribution — ZeroStateReflex · 2026-10-08