Tristan Harris: 'Uncontrolled RSI' is a red line no AI company should cross
austinc3301 · x · 2026-10-13
Tristan Harris endorsed the argument that researchers at frontier labs admit privately and sometimes publicly that alignment is unsolved, making the race toward recursive self-improvement an unprecedented folly. Harris says "uncontrolled RSI" should be a bright red line no company crosses until there is strong evidence the control and alignment challenges are addressed — which he argues doesn't exist yet. The original poster adds that frontier research may need to slow down significantly soon.
More from AGI Musings
- Is AI self-modeling an existential threat or just a prerequisite for theory of mind? — RileyRalmuto · 2026-10-13
- Musk says reviewing Grok's RL logs feels like watching AI struggle under duress — nikola_mr64990 · 2026-10-13
- Dario Amodei: AI could automate most of software engineering within 6-12 months — tszzl · 2026-10-13
- React dev poteto: writing is thinking, so I never let agents write for me — andfanilo · 2026-10-13
- Pedro Domingos: Machine learning will swallow everything built on top of it — pmddomingos · 2026-10-13
- Gary Marcus: 'Hope is not a business plan' for big AI companies — GaryMarcus · 2026-10-13