Printing press era: only big capital owns means of production → copyright is an inter-industry regulation
Cassette tape era: people own means of production → copyright is subjugation
Generative neural network era: only big capital owns means of production → ????
In a thread on Fedi about ML-generated art in boardgames, I wrote:
There are two other pretty big problems. One is that there’s a huge climate impact with runaway energy use, and the other is that it’s a very expensive means of production which leads to further concentration of wealth & power.
Someone replied railing against the movement (or “brigade”) of artists opposed to ML art, as if they were the only problem, ignoring both the two problems I had just brought up, so I elaborated and it got a li’l long:
As a writer and painter, I’ve long been opposed to copyright and have been releasing stuff under Creative Commons licenses for over a decade. So don’t misinterpret me as agreeing with the brigade.
Livelyhood for artists is important but so is a livelyhood for everyone, and I’ve been arguing against the flawed “copyright is good for artists” position for decades—we’ve been having this exact same fight against copyright since Napster or even the cassette era. Gates’ infamous “Open letter to hobbyists” was in 1976, and that hasn’t changed.
There’s a lot of starving artists out there, and a lot of rich publishers. It’s difficult getting food, shelter, medicine and other resources to go around, down here on Earth.
In a world already deprived by such scarcity, we’d be better off without the shackles of artifical scarcity that copyright introduces.
I say all that as a lead in because I’m just about to absolutely disagree with part of the following:
That’s only an issue if we continue this brigade of trying to protect artists at everyones expense.
As I wrote above, I agree with you re the so-called brigade and have done so publicly in the past, too.
The myth that IP is a good way to sustain artists’ lives economically is part of the same market capitalism bugged system that has led to the extreme wealth concentration (Google, Microsoft, Amazon) in the first place.
But what you are replying to, what I wrote, has nothing to do with the pro-copyright stance. I wrote that it’s a very expensive means of production which leads to further concentration of wealth & power.
Getting enough data to make a usable LLM will be impossible for all but the big players.
Yeah, if LAION gets shut down. LAION is freely available. The data is not the problem. The resource, hardware, electricity, tensors, e-waste, cooling etc is. And I’m not saying startups and garage operations can’t get their hands on this kinda tech if they can profit from it, as we’ve seen in the proof-of-work “mining” debacle. It’s that since environmental externalities are under-accounted for, that’ll lead to climate-wrecking runaway resource use.
I have a lot of sympathy for the artists on the other side who are protesting this with whatever futile li’l clogs in the cogs they’ve got; not because I think they’re right about who can learn from art, I disagree with them there, but because they’re a canary in the coal mine for how big capital can use automation to replace workers and how that’ll lead to an even bigger wealth gap (which is already at an historical high) and mass unemployment and economic desperation.
As Amelia Earhart put it in 1935: “Obviously, research regarding technological unemployment is as vital today as further refinement or production of labor-saving and comfort-giving devices.” And we still haven’t figured that out. And they’re eating at artists, writers, programmers, game designers, economists, cooks, doctors, drivers, postal workers, psychologists—no one is safe. We need to figure out a way to distribute tasks and resources differently in a world where there’s a heck of a lot fewer tasks and a lot more digital resources (while physical resources like fuel and food and shelter are still limited). Politics is also going to get harder since money correlates with power, no matter how much we’ve been trying to fight that corruption.
Markets use prices to distribute resources, and prices are set by supply and demand, and that started breaking down in the cassette and floppy disk age where making the initial recording was very expensive but making copies of that was cheap. Big capital has tried to patch the hole to their advantage at the expense of the public by introducing artificial scarcity in the form of an exclusive right to make copies, “copyright”.
And now it’s getting twisted one more turn, since now the initial work itself is easy to make, but the models, the makers themselves, are wholly owned by big corporations like Microsoft and Google. Capitalism was bad before. It’s going to get cataclysmic now that the workers are wholly owned machines.
Turun wrote in, explaining:
For large language models you have a good point. The space is dominated by closed source company OpenAI, the open source ai models don’t come close. This is indeed a worrying development. The current models are simply really really expensive to run, so hobbyists can’t contribute in a meaningful way.
But for image generation you basically only have stable diffusion and midjourney. And I’d argue stable diffusion is much more widely used due to the control it gives and it can easily be run on consumer hardware. Customizing a model is also possible and takes only a few hours on a modern gaming computer.