GPT-3 outperformed human participants on all problem types, both in terms of generative accuracy… and multiple-choice accuracy

→ View original post on X — @aibreakfast