Large Language Models and the Reverse Turing Test