MIRI doubles down: ASI from modern ML this decade is very unlikely to love us
robbensinger · x · 2026-09-14
In the same debate, Rob Bensinger argues the nuances of IABIED chapters 4-5 don't change the core claim: ASI built with modern ML this decade is very unlikely to love us or promote our flourishing. The linked chapter 5 resource page rebuts common optimistic arguments — AI keeping humans around as useful, as 'parents', or as a negligible expense — centering on the point that happy, healthy humans are rarely the most efficient solution to any problem.
Related event: MIRI Researcher and Alignment Researcher Clash Over ASI Doom Arguments(7 posts)→
More from AGI Musings
- Frontier labs reportedly allocate only ~5% of GPU budget to safety and alignment — i_dg23 · 2026-09-14
- Csaba Szepesvári: AI proofs unsettle us just like the four-color theorem once did — CsabaSzepesvari · 2026-09-14
- iamtrask: Follow the Incentives and AI Twitter Makes Sense — iamtrask · 2026-09-14
- RL luminary Szepesvári: researchers actually want AI models that generate insight — CsabaSzepesvari · 2026-09-14
- Pedro Domingos: swap "AI" for "interpolation" and doomer screeds fall apart — pmddomingos · 2026-09-14
- Commentary: frontier AI leaders don't trust each other, so they ask government to chaperone — bilawalsidhu · 2026-09-14