Conversation

I'm as happy as the next person badmouthing OpenAI's product, but what the article on ELIZA beating GPT3.5 (and not GPT4, btw) at the Turing test really illustrates is how useless the Turing test is now that many humans have interacted with chatbots.

1) Only 63% of humans passed the Turing test.

2) Participants said ELIZA was so uncooperative they thought it had to be a human being annoying on purpose and couldn't possibly be a bot – which only happens because they're familiar with ChatGPT.