The main issues with gpt and the Turing test are not that it isn’t smart enough but rather that fine tuning for helpfulness and harmlessness has turned it into

The main issues with gpt and the Turing test are not that it isn’t smart enough but rather that fine tuning for helpfulness and harmlessness has turned it into a Quokka. Completely harmless and naive in its attitudes towards harmful or controversial things, and this is trivially detectable.