Read it if you are scared of AI. It is not as bad as you think. I generally agree with the authors points and also think we are close to the top of the predictive utility of the large languange models. Performance improvements can only develop with improving reasoning and logic follow -- and automatic self-correct, which I do not see clear signs of improving (above and beyond model size.
— Jul 07, 2025 02:13PM
Add a comment
