Richard Socher: goal setting is the key to controlling superintelligent AI
RichardSocher · x · 2026-09-11
In a discussion with nxthompson, Richard Socher argues that goal setting will be central to controlling superintelligent AI: an incomplete objective can send the system down the wrong path, and the smarter the model, the more precise we must be about what we ask it to do. The full conversation on building safe superintelligence is available as video.
More from AGI Musings
- Data center backlash and AI opposition are separate tracks — for now, warns Scholars_Stage — naval · 2026-09-11
- Design Chat at SPC: AI ships code faster, but curation is the designer's job — soleio · 2026-09-11
- Nuclear and bio risks can be addressed separately, and better AI helps build guardrails — robleclerc · 2026-09-11
- Electricians vs. Customer Service: RL Representability Decides Which Jobs AI Replaces — ns123abc · 2026-09-11
- Betting $1M vs $10M that AI won't cause human extinction by 2030 — robleclerc · 2026-09-11
- If we'd built AI in the 1800s, people would've said 'we need ASI to solve diabetes' — zetalyrae · 2026-09-11