Are there medium-to-large-ish, secure Mastodon instances (i.e. ones that aren't going to poop out any time soon) that are trying to resist AI as a matter of policy?
I realize that AI is everywhere and the general attitude is "just give up, you can't win" (so no need to comment that), but are there any instances that are actively trying to avoid integrating with it, as long as possible?
@hollie We have several Rules around AI, including CWs for any mention of AI in any public posts.
But we also have a complete ban on AI generated images.
@hollie Apparently, my stand on AI on our instance is beyond annoying to a lot of academic and industry folks. And they talk about it, and occasionally some of it gets back to me.
They can fuck right off. I'm right, with much thought and knowledge about the issues. And they should have better things to do than complain about how well I run this community instance.
@hollie There has been really amazing and interesting work done that one could currently call "AI".
Sadly, the Big Tech Bro projects have poisoned the whole field and taken over terms like "Generative".
So, we have to require CWs on anything mentioning AI.
Just the other day, some goof was bragging about something he 'vibe coded'. He didn't code anything. He just burned resources. I had to remind him to add a CW.
He didn't see the issue. Ugh.
@hollie I hate having to remind folks to add a CW on this instance. I hate it!!!
But if I don't, our local timeline will devolve into a sloppy swamp.
@kristinHenry I'm in awe that you do this, and am so grateful that you are! I love that you also ban AI art. I'm sorry you've got the folks talking behind your back about it, but yeah, they should have better things to do. And you are right!
Ugh indeed.
@kristinHenry @hollie ban
you really don't want vibe coders in your community. what an annoying crowd
To answer, not that I know of.
Beyond that, being on a medium to large instance doesn't confer any advantage. Spreading our population is better than centralization on the big instances.
@Wintergr33n Depending on the organizational structure and solvency, it does definitely confer an advantage. I can't count the number of small instances I've watched poof out of existence in the last five years; many of those were sudden outages that left their members reeling.
I can count the medium to large-ish ones that have folded. Population size is no guarantee but it is meaningful data when combined with other factors.
I'm not too familiar with any instances that take a firm stance against everything LLM.
mstdn.social has rules against LLM Agents on their server, but content seems allowed. (Surely isn't fav'd or boosted much)
Mastodon server software seems to allow LLM code usage.
https://github.com/mastodon/.github/blob/49b9c64f3e2b6a459e56a05068c05b3247659bb3/AI_POLICY.md
Maybe mastodon.art could fit your criteria. But for some people the strict moderation to keep it a safer space for marginalized people seems to much.
@hollie I'm not sure what your threshold for medium-sized is, but Kind.Social generally has around 1000 monthly active users, several thousand total accounts, and was founded in 2018. We have clearly stated server rules against pushing AI use, creating AI accounts, and posting LLM/genAI output.
While there are certainly some folks who have discussed their dabbles or experiences with AI, and that's acceptable, the majority of users agree that the current "AI" hype and systems are a problem.
Hollie
Sam D
廃墟猫 🐈🌈🐈⬛🐈🐈⬛🐈⬛
Sam Levine
keeps vis.social running
Aria <3 
Brendan Francis O'Connor
Wintergreen
milagemayvary
MurmeltHier
Texan_Reverend
Katzedecimal
AudhdDespiteNoisyAbleism 🇨🇦
Heather 
staringatclouds