I've seen this argument be made a lot in the past year, that "Oh these things are just tools and you are an idiot for having a critical view on them". I've been thinking a lot about this and I've come to the conclusion that it's based on a flawed understanding of economics, and of quality generally.
Do we measure software engineers in terms of their lines of code per day? I feel that this is basically what you and others are doing when you compare writing or illustration to digging holes. You're seriously misunderstanding the economy of these fields. Productivity has almost nothing to do with the market for art or writing. What matters is the quality of a work (very hard to measure and deeply ingrained in our psyche) and the attention a work receives (Software engineers should be more familiar with this. It is the "algorithm" people love to disparage). We had more than enough productivity already in these spaces. Does giving people a content firehose actually do anything to these markets? Have you read any books written using AI? Do you know and follow any AI artists? Is anyone besides Nvidia and OpenAI making money here?
I understand the appeal of the models, and as a software engineer I've gotten at least some use out of them. But even in that space I have to wonder if producing more code is a net positive. There's a lot of drawbacks to producing more code. There's more chances to write bugs, more you have to document and maintain, and you're potentially overlooking simpler and more elegant solutions.
I guess all I'm trying to say is when you say this is just a hole digger, you're glossing over a lot of nuance.
> Do we measure software engineers in terms of their lines of code per day? I feel that this is basically what you and others are doing when you compare writing or illustration to digging holes. You're seriously misunderstanding the economy of these fields. Productivity has almost nothing to do with the market for art or writing.
Depends very highly on what kind of art we're talking about.
There's the "I get my stuff into galleries" artist. This IMO has an extremely high social component. It's in good part about who you know, how well you can spin your work, how much some rich guy has paid for it in the past, etc.
There's the "I sell commissions on Deviant Art" artist. That's to me a kind of hole digging position already. Certainly you need to produce good work, but what seems to make a good artist in that scope is good work, reliably, fast.
There's the "I illustrate stuff" artist. A more properly commercial version of the above. From my limited experience there's demand to pump out work quickly and as cheaply as possible. Eg, a book may need a cover or a few illustrations, or an ad might need something drawn on it. In many cases the client is basically looking for a hole to fill. The cases where somebody wants to have something truly beautiful that lots of effort went into seem vanishingly rare. There's a reason why corporate styles like "Alegria" exist and are annoyingly common.
There's the "I draw comics" artist. Those need extreme consistency because they're on a tight deadline.
> Do you know and follow any AI artists?
Honestly, I'm beginning to. AI is just a tool like any other. Some people use it to faff about and make random stuff, but I'm starting to notice people using it to produce things that have a coherent plot or tell a story, or at least keep a consistent setting. It's still janky at this point because it's hard to maintain consistency.
Really, there's nothing weird about that. Think about supremely lazy things like Garfield or Ctrl+Alt+Del. They still had a following even when graphically they were supremely lazy. If you can copy/paste body parts to assemble a comic and still get some people interested, no reason why you couldn't do it with AI. Heck, it'd come out looking better at this point.
Do we measure software engineers in terms of their lines of code per day? I feel that this is basically what you and others are doing when you compare writing or illustration to digging holes. You're seriously misunderstanding the economy of these fields. Productivity has almost nothing to do with the market for art or writing. What matters is the quality of a work (very hard to measure and deeply ingrained in our psyche) and the attention a work receives (Software engineers should be more familiar with this. It is the "algorithm" people love to disparage). We had more than enough productivity already in these spaces. Does giving people a content firehose actually do anything to these markets? Have you read any books written using AI? Do you know and follow any AI artists? Is anyone besides Nvidia and OpenAI making money here?
I understand the appeal of the models, and as a software engineer I've gotten at least some use out of them. But even in that space I have to wonder if producing more code is a net positive. There's a lot of drawbacks to producing more code. There's more chances to write bugs, more you have to document and maintain, and you're potentially overlooking simpler and more elegant solutions.
I guess all I'm trying to say is when you say this is just a hole digger, you're glossing over a lot of nuance.