Softmax's multi-agent RL AGI path sparks AI safety community split
repligate · x · 2026-09-13
@BjarturTomas calls Emmett Shear a 'pessimizer', arguing his company Softmax is pursuing one of the most dangerous paths to AGI — multi-agent reinforcement learning in competitive environments — while mistaking it for safety research due to confusions about evolutionary history. repligate pushes back: not bullish on Softmax, but exploring the most dangerous-yet-inevitable directions at small, harmless scales is actually good for safety.
Related event: Softmax under fire for pursuing the most dangerous AGI path(2 posts)→
More from AGI Musings
- Prediction: OpenAI's robotics 'ChatGPT moment' lands next year with a browser-prompted robot demo — flowersslop · 2026-09-13
- AI's challenge to academia's self-serving prestige hierarchy draws sharp debate — soumitrashukla9 · 2026-09-13
- Stanford HELM researchers compile reading list on LLM sycophancy and AI social harms — chrmanning · 2026-09-13
- Chris Manning proposes Stanford NLP as independent AI alignment evaluator under Amodei's plan — chrmanning · 2026-09-13
- Top AI leaders unite to warn technology is advancing too fast — KatieMc___ · 2026-09-13
- If every field gets its Stockfish moment, what does ideal education look like? — voooooogel · 2026-09-13