Sycophancy to subterfuge: Investigating reward tampering in language models (www.anthropic.com)
![](https://kbin.burggit.moe/media/cache/resolve/entry_thumb/2e/0e/2e0edcd1b3ee980e8391fcc06fabc0928bb4010d468550e37e58a9bab77e0a90.jpg)
This magazine is from a federated server and may be incomplete. Browse more on the original instance.
Many people have been surprised how quickly open-source AI has kept pace with the AI efforts getting billions in investor funding. It’s worth wondering if the same may happen with robotics. After all, robotics are primarily AI too, though embodied in a 3D environment. Recently two major Chinese manufacturers, UBTech Robotics...
cross-posted from: lazysoci.al/post/14640253...