Doing Precise and *Reliable* Automated Reasoning like It is 1972

I plan to show how current state of the art LLMs cannot really solve high-level logic puzzles that require precise rigorous logical thinking. Showcase example: Einstein’s puzzle
However, AI systems in the symbolic reasoning branch of AI and dating back more than 50 years ago, such as Prolog can do this almost effortlessly.