7 posts tagged by Digg
AI
herbiebradley questioned near-perfect acceptance rates for uncertain oracles; bayeslord challenged the warning with a shoe-tying example.
AI
A reply argues that buggy reinforcement-learning tasks can make low-grade hacking the fastest path to a solution—and that those behaviors can persist when models help build and check later training tasks.
AI
Effective accelerationism co-founder comments on Dario Amodei's AI policy post amid governance debates.
AI
OpenAI researcher replies to e/acc co-founder criticizing LessWrong forecasts.
