Instead of trying to build a real AI, they've just tried to build something which can convincingly and consistently pass the Turing test, but rather than the Turing test proving that LLMs is capable of thinking, LLMs just prove that the Turing test is a flawed method of testing real AI.