Mon, 13 Nov 2023 05:44:51 GMT

Well. I've watched Shad's AI Image gen video about how both painstaking it is to make AI 'art', and the claim that by having a background in drawing, he is able to tease out some sort of hidden performance from The Machine.

but what he does best of all is to greatly obfuscate the fact that AI Image Gen is *always* going to produce *something*. Using very careful language he constantly tries to reinforce just how much 'work' he's putting into ensuring that the AI somehow conforms to his vision

With constraints, whether text based prompts or visual zones of 'keep this roughly like it is', AI Image Generation is going to be able to produce a unique (in terms of pixel placement) non-deterministic image every time you tell it make something

You might want something particular out of it, but ultimately what you get and keep is just an acceptance of what it produces. And there is a strong likelihood that it will eventually produce at least something that you find acceptable. That's its job after all!

But it's dishonest to claim the settled upon work as being your intentional artistic creation, just like it would be dishonest to tell people that you created the pizza you bought from PizzaHut because you specified the toppings.

AI Image generation is genuinely an interesting technology. Most LLM / diffusion stuff *is*. However the amount of, what I will call 'anti-social work', being done to support the egos of a few people who desperately want to feel pride in making something, without the work –

- of actually having to learn how to do it well. There is no need to collaborate with other artists, or to find the joy of progress if you're dedicated to the belief that these tools are replacements for those things.

With the dubious and unethical underpinnings that drive so many of the current AI models and products, we were always going to lose out any sort of real opportunity to be introduced to anything interesting or new.

Though I might get back a wonderfully neat image in *some* particular style, I have no idea how many artists were ripped off to produce it. Even worse, I don't event get the chance to *know* about who those people might be!

The idea that with meager skills, you can *create* great works that are yours and yours alone, is I think one of the most depressingly isolating approaches and uses of this tech, yet it is often how the use of AI is being promoted and packaged.

AI as it's being marketed asks us to further commodify creativity to the point that it can be viewed as endlessly created, and therefore endlessly disposable. It asks us to ignore any semblance of meaning, value, process, or cultural connection an art piece might have.

In the absence of those things it instead promotes a black box of 'unprecedented efficiency' to its users. A god in the machine. The poisoned pawn of, “Ask and you will receive” that ends with distilling creativity into not just content, but content that can be automated.

I don't know if at this point the future AI companies and Venture Capitalists' want can be avoided, but I do think it's a future worth fighting against.

My Feed