Humans will probably always set AI's highest-level goals, however smart AI gets
burny_tech · x · 2026-11-11
The author argues that no matter how intelligent AIs become, some humans will likely always retain the authority to set their highest-level goals — a brief take on the boundary between human control and AI autonomy in an AGI future.
More from AGI Musings
- PhD unemployment is just 1.5%, economist notes while listing four moral demands for grad programs — Afinetheorem · 2026-11-11
- Economist pushes back: we train PhDs as a social good, training costs aren't an accounting fiction — Afinetheorem · 2026-11-11
- Ex-OpenAI dev: 1.2B use ChatGPT, and AI will integrate itself, removing the diffusion bottleneck — willdepue · 2026-11-11
- Eric Topol interviews Ethan Mollick on his new book CO-EXISTENCE and where AI is headed — EricTopol · 2026-11-11
- Peter Diamandis: Chores and caregiving beats dancing-robot videos as a robotics roadmap — PeterDiamandis · 2026-11-11
- OpenAI's Will Depue: all non-artisanal human economic labor is "totally fucking over" — willdepue · 2026-11-11