you are viewing a single comment's thread
view the rest of the comments
[–] 2 points 1 month ago

AI safety needs to be treated with defense in depth. From silicon to training data to neural architecture to RL tuning.

  • source