Verily, hell hath frozen over!!!

#CriticGPT hath bin bestowed upon us!
Rejoice, rejoice greatly!

https://openai.com/index/finding-gpt4s-mistakes-with-gpt-4/

/cc @tante @isotopp

@alexshendi @tante @isotopp So it "outperforms" in just over half of the cases, but how do the rest of the cases impact that (where they have to review more data first)?
How is the *overall* productivity?