Probably nothing, but at least the people running our instances aren’t directly involved in doing it to us. The people behind this community aren’t pretending that it’s OK to exploit other people who are just trying to connect with each other. And even if OpenAI still slurps up our data, Sam Altman and the other AI billionaires don’t make as much money off of us because they don’t own Lemmy.world and don’t profit off of any deals or ads here. These things make a big difference to me.

what

yeah, but the protein folding AI is a totally different kind of technology from what OP is talking about. AlphaFold has nothing to do with Ollama or whatever.

vaguely related: I kinda think society should scrap the term “AI.” I mean, my washing machine has an “AI” setting that… senses the amount of laundry in the load? Lol. Sure, we’ll call that AI. “AI” is such a mess of a term that it’s practically meaningless.

Ethical training data could exit, but I don’t think public domain/creative commons/licensed are quite good enough (almost no one who contributed knew what AI would become until recently, and some of those people are still alive). Although that’s better.

A truly ethical training dataset for AI would be one that everyone who created the training data opted into. And I don’t mean their employer or photography sharing platform signed a license deal, I mean they personally opted in. If AI is so great, I’m sure plenty of people will be happy to contribute their art, photos, writing, code, social media posts, etc. All the pro-AI people can just band together and make their own dataset to use.

That doesn’t solve all the other issues with generative AI, but it would make a huge difference.

Sam Altman partially owns it. Reddit is selling every user’s comments directly to OpenAI with no way for them to opt out of AI training. 'Nuff said.