Comment on Do you think Google execs keep a secret un-enshittified version of their search engine and LLM?
krooklochurm@lemmy.ca 18 hours agoI started tinkering with ai right around the time ChatGPT rose to prominence. Locally. On my own machine.
I’m not a doctoral level researcher but I mostly get the tech.
I couldn’t agree more. People use ai as a blanket term and don’t understand the difference between an LLM and GAN or any of the dozens of other kinds of models.
If it’s ai it’s bad. Just full stop. Like. The anger of people decrying the death of artistic beauty on subs that prominently feature ms paint stick figure drawings and shitty distorted images makes no sense to me. This isn’t costing anyone’s job. It’s fucking garbage content, with no agenda, and always was.
Having autonomous LLMs posting things is problematic but have ai generated shitposts isn’t.
There is fuck all wrong with using ai to make art to hang on your walls, or funny t shirts, or ridiculous banners, or funny pictures to share with friends. The people that decry the death of art have never bought anything in a gallery, they were fine with artists getting paid fuck all before ai. They weren’t contributing to artists’ living in any meaningful way.
And like. The most vocal critics seem to understand the least about it. Such that they hate it because it’s made with ai just assume that someone’s made it using OpenAI because that’s the only thing their rage-addled minds can process existing.
They say it’s theft and we should ban everything (how’s that working out for you?) instead of clamouring for fair compensation for anyone whose work is being used to train a model.
They’ll yell: all these models are based on theft. And sure. But a) I don’t give a flying fuck about a corporation’a right to exploit an artist and profit off their work and never have. And b) will respond to the suggestion that we create new models that fairly compensate people by yelling louder and becoming irate.
They’re not rational. There are many valid criticisms of the tech, but you can’t even talk to these people addressing them. They won’t hear it.
BCOVertigo@lemmy.world 15 hours ago
I’ve commissioned paid art for rpg campaigns, and I can’t draw a distinction between AI and LLMs because I get yelled at by people saying, “Its just the name of the field! Nobody thinks the Sims games are actually intelligent!”
So am I allowed to draw that line now? And do you see that me using a comfyui on my local machine does actually mean an artist won’t get paid? This position isn’t 100% strawman.
My main issue is that I think maybe they can’t be patched because they’re not deterministic systems, and I have personally been asked by an executive whether a team could reasonably be replaced by LLM agents behind their backs as soon as the tech was available. How was I supposed to form your opinion given those experiences and why do you think that I’m rage addled rather than just tired?
krooklochurm@lemmy.ca 15 hours ago
I don’t really understand what you’re getting at here.
BCOVertigo@lemmy.world 13 hours ago
Sure, I’ll line up our talking points with a keyboard and elaborate.
When I say “These models aren’t actually intelligent, talk about specific technologies and models without lumping them all together under AI.” I am met with scoffing saying that everyone knows AI is just the name of the field so I should stop being pedantic. Now you’re painting the opposite picture that a person like me lumps them together because of ignorance, and that’s frustrating. Maybe some of those assholes you encountered are exhausted because the artificial cheers for this technology are inescapable and that’s coloring their interactions with you. I don’t feel like it’s a fair point to call people ignorant for having to wade through manipulative marketing language to have a discussion. Hopefully you can sympathize with that a little.
I’ve paid artists for art related to tabletop RPG campaigns in the past (some but not all). Now everyone shows up to games with genai character portraits and I believe my artist is going to have to quit or at least scale back because of lack of demand. I would ask that we keep in mind how gig work, commission, and subscription models form the basis of most modern artist’s income. Anecdotes aren’t data, but data doesn’t give you an email thanking you for your support while giving up on its dreams. The taste in my mouth just didn’t go away when genai invaded shitposting too. Does that make some sense?
I don’t think the tech is being used well by the largest actors wielding it. I don’t think genai can be secured or patched in the traditional sense because its not deterministic, so prompt diddling will always be a cat and mouse game. I don’t think that hallucinations will ever be meaningfully solved. I’d love to learn that the overall adversarial success rate has lowered from 95% since the study I read last year! It’s not that I won’t hear solutions, I just have very little confidence in them given how this saga has developed over time.
Please spare a little charity for how rattling this shit is to regular people. I believe you that you’ve talked to some absolute dipshits, and I hope you get fewer in the future. I also hope that maybe you can reinterpret some of that vitriol in light of how these systems might have resulted in them being laid off or otherwise maligned. Maybe it’s not that irrational even if they can’t articulate their feelings well. Personally, I’m stifling irritation at an executive who asked me to replace a technical team with an agentic framework (behind their back) despite the technology being wildly inadequate. I hope none of that overflows onto you.