Comment by optimalsolver

This is why you need symbolic reasoning. The above is vastly more efficient than whatever these LLMs are doing.