This question:
Content removed. I do not consent to "AI" scraping
Screenshot at the time I wrote this:
On other SE sites as well:
Rolling this back will return in a tug of war.
What to do?
Possibly relevant: Who owns the content I post?
1
-
This person went through all their SO sites and did this to all the questions and answers they've made. It appears all have been rolled back.– music2myearCommented May 7 at 18:13
Add a comment
|
2 Answers
Flag the questions with a custom flag. We get automatic notifications on excessive vandalism but this is probably something we want to keep an eye on.
As for what's happening...
Stack Exchange/Stack Overflow has been tying up with genAI companies to sell access to an API. We're still waiting on clarity on what this'll look like, but they're promising attribution to our posts and such.
nixCraft on Mastodon has been encouraging folks to stop contributing to SO and network sites, including deleting posts. Vandalising your posts is against the TOS, and frankly hurts the human/open web more than it hurts genAI companies who probably have scraped SE or sites scraping SE already - and practically vandalism and deleting your account the regular way does not render your content inaccessible to SE should they wish to anyway.
It also means, well it's the community who needs to deal with the fallout of someone who isn't active on the network messing with things, just like the glory days of twitter, with little actual effect on the people they want to hurt.
If there's a need, we may suspend users to keep them from vandalising their posts, and we will definitely moderator-message users as needed.
answered May 7 at 11:13
-
7
they're promising attributionthat is not a promise they can make, because of how LLMs work. I googled it and I was unable to find a source for the claim that they promised that. Do you happen to have a link? OpenAI did promise to give Stack Overflow attribution, but not the person who actually posted the content.– GantendoCommented May 8 at 2:25 -
1Note also that Stack Exchange has shown it has little to no respect for licenses. meta.stackexchange.com/questions/333089/… And large AI companies have not asked for permission to use the content they've used to built the models.– GantendoCommented May 8 at 2:35
-
This might be a question for main meta - we don't really have the full details on implementation of these things yet.– Journeyman Geek ModCommented May 8 at 3:07
-
1Ah perhaps you were referring to meta.stackexchange.com/a/399630/1507004 ? LLMs and attribution are fundamentally incompatible, and they clearly have no idea what they are talking about.– GantendoCommented May 8 at 20:11
-
@Gantendo: LLMs as implemented today may not be capable of attribution, but that is not a fundamental limitation of LLMs (It is very deeply embedded in the structure presently being used, there definitely is no quick fix. But we don't know how much work has been put into competing structures, or how close to maturity they could be.) Commented May 14 at 15:04
-
And for your "StackExchange has no respect for its license obligations" this was a far better example: meta.stackexchange.com/questions/271080/… Commented May 14 at 15:07
-
@BenVoigt There are probably many better examples of SE ignoring licenses, but I am no SE-expert. As an immortal timetraveller I am aware of all past, current and future tech.– GantendoCommented May 14 at 15:33
From Chat (thanks https://superuser.com/users/1210833/notthedr01ds) there are other profiles editing and deleting their answers.
It may be worthwhile having some sort of flag for times an answer is edited significantly and "AI" is in the edit.
Evidence
- Profile https://superuser.com/users/905389/ai-test-account Deleted answer: https://superuser.com/a/1589527/74026
