Hm, I think every geek made a chatting computer from Acornsoft's Speech or DECtalk or whatever the Amiga and Atari equivalents were in the '80s, but it wasn't so logorrheic or self-confidently wrong as ChatGPT. It's really hard to have a nice conversation with ChatGPT - like if you want to let it out then a much dumber interface that's closer to Eliza is more helpful, and if you need a question answered then a search engine + existing human responses are way more useful. I hope we get to the point with ChatGPT that we already seem to be at with DALL-E, where we say "oh that's fun" but understand that it's not thinking anything like a human and so is more fun and useful in very specific cases than as a general intelligence.

I feel like ChatGPT's basic vision is to simulate a mediocre software engineer who thinks they're a genius in every field, and we have millions of them already. In this sense maybe OpenAI is performative art of Genesis God making human life in His image.