Because what we really need is a standardized benchmark to tell us which AI model is "best" at predicting the meaning of a document, without questioning the validity of those https://www.reddit.com/user/shhdwi