The new owners have been trying very hard with the "if you can't beat 'em, join 'em" approach. They know this is radically against community consensus (it's been shown to them on the meta site over and over) - so they just get sneakier about it.

Notably, after getting completely humiliated with https://meta.stackoverflow.com/questions/425081 in June 2023 (right after a moderator strike had just started, protesting the staff trying to prevent them from removing AI content from the site), and getting embarrassing feedback on the feature (https://meta.stackoverflow.com/questions/425162), they came back last November with https://meta.stackoverflow.com/questions/432154 and have been forcing it through.

Where's the AI of tomorrow going to learn from if nobody is posting Q&As online anymore?

From the official docs? AI is good at sumarizing after all

I've been trying to get ChatGPT to write some Emacs Lisp for me and it sucks. Few things are better documented than Emacs. There's several hundreds of pages of documentation, and several million lines of Elisp, but apparently that's not enough.

And I'm not asking for some beautiful architecture from ChatGPT, I'm just asking for simple hacks that get the job done. Elisp is designed to make simple hacks easy, but not easy enough for ChatGPT I guess.

Like, I asked it to make a command which would move the "mark" and the "point" so that full lines were selected. If a selection covered only part of a line, I wanted the selection to expand to cover the full line. To do this, all you have to do is move the mark and the point so that they surround a line. ChatGPT couldn't do it. It would only move the point, but not the mark, it wouldn't do anything with the mark. I explicitly told ChatGPT "no, you have to set both the point and the mark correctly", and then it wrote even more code that only adjusts the point, but not the mark--it would move the point to the beginning of the line, and then move the point to the end of the line--stupid.

Good luck with that. The last thing I used SO for was getting answers for SwiftUI and I can assure you the official docs did not contain the needed information.

That’s a tomorrow problem.

(As someone who is all too often hired tomorrow, at a fraction of the before rates, to clean up this mess)

information will be repackaged like credit default swaps in the mid 2000s.

Other people's code on GitHub.

The other people's code which will have been AI generated by older, dumber models in many if not most cases? Possibly even written and committed by AI agents with no human review at all? European royalty tried this kind of thing and it didn't end well.

You got this exactly right. I've been trying to sound the horn on this issue for years now, since GPT-3 got released and exploded into the mainstream and you could read and hear "you don't need to {x} anymore with AI" all over the place.

AI just stole all its content, I wonder if they will choose to sue.

Stack overflow never owned the content, it is and was Creative Commons: https://stackoverflow.com/help/licensing

That license is granted to the community, but per the linked Terms of Service, the company gets a slightly different license (https://stackoverflow.com/legal/terms-of-service/public#lice... ; scroll to "Subscriber Content"). Emphasis on:

> ... and you grant Stack Overflow the perpetual and irrevocable right and license to ...

That license comes with obligations that the AI companies aren't following.

Its highly debatable if that's true.

The cc lience only applies to the presentation of the information, it doesn't apply to the factual content of the information. Which part openAI used is a matter of much controversy.

Cursor once wrote a comment, I prompted then « what is the source of the comment » and it replaced the comment with a stackoverflow url in which the page contained the said comment verbatim. I didn’t expect cursor to paste the full url

Well.. if given the right prompt.. it will effectively fully reproduce a stack overflow post in it's original form then there is no controversy and we can all see plainly what's going on behind the scenes.

Will it actually? There have certainly been incidents of that sort of thing, but from what i understand people have not been able to reliably make that happen for specific posts .

Define "it."

And yes. I've gotten many free LLM tools and demos to fully recreate stack overflow posts or blog articles. They seem to have a habit of copying comments verbatim which is usually good token to search for to find the original "inspiration."

It pretty reliably does this. The simpler the program you ask for the more likely it is to just copy one. Which we can argue that the simpler the program the fewer the plausible implementations but when it copies the comments so exactly and positions them identically then there aren't any other conclusions to reach.

The current set of "AI" companies are just in the business of whitewashing copyright violations.

Sure, but copying 1 or 2 sentences verbatim into a long program probably isn't enough to win a copyright suit (de minimis). Or at least it would be pretty grey. Similarly, the simpler the program, the less likely the creative intent shines through.

(Mostly im just trying to say these things depend on the details, and the anti-AI crowd tend to handwave the details away. They might still be right, but its far from an obvious slam dunk)

A distinction that is unfortunately as important as it is meaningless