Of course LLMs have limitations. So do humans. But LLMs clearly engage in reasoning. I think denying that at this point is just crazy.