I don’t count productivity or coding as a net good for what it’s worth. Fun, sure, but it’s truly incredible how the influx of AI hasn’t been used to create a more efficient way to distribute resources irl
Tech bros are a scourge on humanity and I truly wish for nothing but the worst for them
it has practical applications but like 90% of them are not in the fields it is being used in. we could train lumberjack robots and automated mining drones with this shit but that's hard so we just decided to steal art and automate one of the few things almost all humans actually enjoy doing (yes there are people who enjoy cutting wood and mining but at least they're not the majority of humanity)
honestly im not sure if automating labor in general is worth it. something about labor metabolizing human's relationship with nature or something. and i say this as someone who hates work.
it would be great as a way to basically "cover" for disabled people though, allowing us to be a primarily non-automated society that also doesn't have to make decisions about what sick or disabled people to care for (we could care for all of them)
edit: i changed my mind again (though the thing about getting resources to take care of all disabled people is still true ofc), I think automation is good but it needs to be designed by and controlled by the people who enjoy doing the things it's automating.
This way it would function as an extension of people, like IRL qol mods, rather than a replacement, so instead of getting rid of labor which many people actively find fulfillment in, we just remove the parts people don't like doing, and what those parts are would be decided democratically (and in many places not even universally, so what would be considered reasonable to automate would change from place to place etc, allowing people who want to do those things to just move somewhere else to a community that vibes with them more).
The way actual AI art functions now is as an extension of human artists, not a replacement. You can't just wind up the algo and watch it go, not only does it require curation and inpainting and touch ups, but it also requires an artistic vision that can only come from an artist. Without an artist it just churns out bland content, inherently derivative and filled with errors.
My only problem with AI-assisted art is that people who don't have any artistic taste, like managers, think that it's good enough to replace artists. That's not the technology's fault, though.
Yeah except… the latter thing you mentioned is what companies want, because they care more about profit than producing things.
Plus the “always just a recreation of existing cultural perspectives” thing is true too. There isn’t really a way around that in it’s current form. I think any tech nerds who are lefty should have the goal to implement more ways for artists to reinterpret and include their own views in what the AI is creating
I don't see why an AI couldn't be trained on specifically non-Western/counter-cultural/marginalized art. It seems trivial?
The bigger problem I see is that it can never make anything truly new, which is why it's important for artists to keep making new art. I just think that AI, as a tool used carefully by artists to enhance their work or to take shortcuts on repetitive work, is fine.
As you said, companies are happy with AI goop that's "good enough" and do not care about art, but that's not the technologies fault.
EDIT Well wait, no, actually the way the plagiarism machine seeks to reduce all artwork into data points is also kind of bad since there's no mechanism to force these AI companies to actually compensate artists.
Lol I just realized you said exactly what I said with less words, RIP me. Sorry comrade
Sorry, let me clarify. The issue isn’t the art it’s being trained on, but the fact that the artist can’t (currently) input their own personal opinions and thoughts. You can train it on countercultural art, but you can’t, like, specifically depict every aspect of how you personally view that art’s depiction of, say, a cat, so it sort of stays “stuck” on the level of simply depicting the average cat in the art it was trained on. It can’t actually go through the art cycle of interpretation -> reinterpretation -> etc etc on the granular level traditional art can, instead being limited on only doing this on the abstract level of the prompts (which aren’t the end product anyways, so doing reinterpretations of them based on your own opinion is counterintuitive at best and at worst will not result in any sort of reinterpretation in the final product at best).
This is all a way for me to overexplain the fact that it cannot accept granular input based on the user’s subjective experience. This a huge limitation for it and, I would be willing to bet that if we fixed it, it would stop looking as much like art generators and more like how custom music synthesizer creation works or something. The AI would stop being an artist and start being a brush, basically.
we've always been able to do that, we don't need the power of A.I. to automate even more labor