Emmett Shear's Softmax slammed for framing dangerous multi-agent RL as safety research
BjarturTomas · x · 2026-09-12
@BjarturTomas calls Emmett Shear an interesting example of a 'pessimizer', claiming his company Softmax is working on one of the most dangerous possible paths to AGI — multi-agent reinforcement learning in competitive environments/ecologies — while mistaking it for safety research due to basic confusions about evolutionary history. This is the original tweet behind repligate's rebuttal.
Related event: Softmax under fire for pursuing the most dangerous AGI path(2 posts)→
More from AGI Musings
- Blogger: Dario's AI safety essay reads like US-style military-civil fusion in disguise — bilawalsidhu · 2026-09-13
- Peter Wildeford's case for pacing AI: 'already low-key out of control' amid an 'insane September' — KatjaGrace · 2026-09-13
- Altman says top AI leaders meeting to slow down AI 'will happen' and may be underway — 4KTV · 2026-09-13
- Beff Jezos: basically all frontier labs already have pseudo-recursive self-improvement — beffjezos · 2026-09-13
- 'Independent evaluators' will be the most powerful people on the planet — cgarciae88 · 2026-09-13
- whurley calls AI safety push regulatory capture to kill open weights — whurley · 2026-09-13