> There is no theoretical or empirical basis for LLMs having any reasoning capability. Yes there is. Learning to predict the next token implies a lot of thing

> There is no theoretical or empirical basis for LLMs having any reasoning capability. Yes there is. Learning to predict the next token implies a lot of things, among which is also logical reasoning. The chain-of-thought approach shows that when you stimulate this behavior, you get higher accuracies.