Originally, from here[1]. Large Language Models seem to be better than humans at text prediction (at least on their trained data set). So it seems like the limi

Originally, from here[1]. Large Language Models seem to be better than humans at text prediction (at least on their trained data set). So it seems like the limiting factor for AI is more the goals we give it rather than the data or the size of models, at least right now. [1]: https://www.lesswrong.com/posts/htrZrxduciZ5QaCjw/language-models-seem-to-be-much-better-than-humans-at-next https://www.lesswrong.com/posts/htrZrxduciZ5QaCjw/language-m...