Raw LLM Responses

I think "chatbot" style AI's are really freaky to some people, but I don't understand why the Turing Test is thought of as a valid test of sentience. The Turing Test is pretty literally a test of whether a chatbot can fool a human, and that's been pretty easy for a while (especially chatbots that are trained to simulate emotions, or be rude). Of course a chatbot trained on human content will give human answers. It's going to say "I'm afraid of death", because humans write about being afraid of death. It takes human-written inputs, and spits out an output. In other words, building a machine that takes sentences and provides responses isn't evidence of any internal state, sentience, or emotion.