Absolutely amazing stuff. I am now three scores and ten in my life time, seen a lot of changes from slide rules->very fast to calculators->very fast to pcs, from dot matrix printers to lazer jets and dozens of other things. Wish AI was available when I was doing my PhD. If you know its limitations it can be very useful. At present I occasionally use it to translate references from wikipedia articles to bibtex format. It is very good at this, I only need to fix a few minor errors, letting me focus to the core of what I am doing. But human nature always resists change, especially if it leads to the unknown. I must admit that I think AI will bring negative consequences as it will be misused by politicians and the military, they need to be "regulated" not the AI.

Yeah, they made something that passes a Turing test, and people on HN of all places hate it? What happened to this place? It's like the number one thing people hate around here now is another man's success.

I won't ignore them. I'll continue to loudly disagree with the losers and proudly collect downvotes from them knowing I got under their skin.

Eliza effectively passed Turing tests. I think you gotta do a little better than that, and 'ha ha I made you mad' isn't actually the best defense of your position.

Eliza did not pass Turing tests in any reasonable capacity. It took anyone 10 seconds to realize what it was doing; no one was fooled by it. The comparison to modern LLMs is preposterous.

GP doesn't have to defend their position. They like something, and they don't shut up about it even though it makes a bunch of haters mad. That's good; no defense required. On the contrary: those who get mad need to defend themselves.

I mean even on the Wikipedia article it says: "Some of ELIZA's responses were so convincing that Weizenbaum and several others have anecdotes of users becoming emotionally attached to the program, occasionally forgetting that they were conversing with a computer.[3] Weizenbaum's own secretary reportedly asked Weizenbaum to leave the room so that she and ELIZA could have a real conversation. Weizenbaum was surprised by this, later writing: 'I had not realized ... that extremely short exposures to a relatively simple computer program could induce powerful delusional thinking in quite normal people.'"

That's why I said it passed the Turing test. It did, in the wild. I think if you set up an adversarial Turing test it'd fare much more poorly, but consider this: anyone dealing with it knew they were dealing with a computer, not just 'text from a mystery messager in another room'. So in a sense it more than passed the Turing test, in that people started out knowing it was not a person, yet began to treat it as one anyhow.

This is more about how little it takes to pass a Turing test and be treated as intelligence… which is mighty salient to discussions of what we now call 'AI'. Some of you are taking today's Eliza mighty seriously.

I must have been about 8 years old when I played Eliza for the first time on some home computer. Even to a dumb kid it was very obvious how it worked, two questions in. And I am far from being a genius. Maybe it was more convincing in English than in my native language, but "delusional thinking" seems about right.

Tbh, I don't care about Turing tests, or the AI narrative of big corps or doomers. I do care about my personal experience with LLMs and diffusion models. Not in a mighty serious way, but in mighty fun and entertaining way that wasn't possible before.