Researcher Optimistic on Alignment, Calls for Lab Support

tszzl argued that solving alignment with verifiable subgoals may be no harder than Millennium Prize problems, praised external teams like Goodfire, and called on labs to provide compute and models to alignment researchers.

2026-09-15 ~ 2026-09-15 · 3 related posts

• tszzl: Alignment May Be a Grand Challenge, But Who Says It's Harder Than the Millennium Problems? — tszzl · 2026-09-15
• tszzl Backs External Alignment Labs: Lab Pressure Cookers Crowds Out Deep Thinking — tszzl · 2026-09-15
• AI Researcher Urges People to Join Alignment Work, Citing Lab Compute Access and Independent Labs — tszzl · 2026-09-15