Generally, we see that our participants start out with far worse accuracy on these tasks than our models do, and that by interacting with the models, they’re able do about 10 percentage points better.

→ View original post on X — @anthropicai,