Downloading artists' songs is not equivalent to people using AI to generate sounds/images and claiming they are an artist. I really don't have a problem with people using AI to generate sounds/images for their own personal enjoyment, but taking what it generates and then telling others that you "made this" or are an artist is deception.

> artists shouldn’t expect to be paid for their digital output

The issue is the notion that an artist gets to control what one does with their personal property that isn't the artist's property. No one is saying artists shouldn't get paid. Artists should get paid but setting up a system that surveils everything I hear and see to enforce it is too much.

> Many people just don’t care about this stuff

I agree with this though I don't follow your tie-in to piracy. Most people do not really care about music, and the industry has known this and delivers most music through ad-supported channels and shapes what music production it can to fit this. The ugly truth is that there's probably a lot of people who wouldn't mind listening to AI radio, it's probably coming, and it will be good enough that a sizeable percentage of the population will enjoy it and not care.

The real art has always been outside of the industry though, and that won't change in the AI age.

> people using AI to generate sounds/images and claiming they are an artist.

> but taking what it generates and then telling others that you "made this" or are an artist is deception.

I think there's a fundamental problem with this line of argument, which is that it's naive elitism that's been historically leveraged since the 20th century at minimum. "Your music was made on a computer? That's not real music! All you do is push some buttons!" with equivalencies for all mediums of digital art.

There's absolutely a reductionism I think that happens with naysaying generative AI, where it's almost a kind of naive agreement with the VC hypetrain. There's a belief that effective usage of AI is just dumping 20 words into a text input and the maximal quality it's capable of gets spit out 5 seconds later. It's the same variety of reductionism as above where ignorance of how wide the domain is precludes the ability to understand where the skill ceiling is (or what is even involved). The idea that this isn't even the tip of the iceberg nets an expectation that "advanced" AI usage extends to random prompt engineering at most.

Problematically, "AI generated" doesn't really say much about the process, intent, or really much of anything beyond a specific computational architecture being involved at some point of the workflow. It could mean they slapped a tweet-length stream of consciousness into a shitty web prompt, or it could mean constructing a complex pipeline, combining processing steps and multiple models like hotpatching synths, using an intuition built upon thousands of hours of learning their tools, using fine-tuning adaptations that they spent even more time building, and carefully analyzing and responding to the end result. In the former case, I don't think there's much artistry involved and it exclusively produces crap. I think it's a massive disservice to the latter however, to suggest it doesn't involve extensive artistry, or that the paradigm of the tools themselves strip the notion of art away.

I understand wholly why the first case gets primacy, but if we're to philosophize about these tools then I think we have to fully account for the second case.