Even outside of capitalist exploitation, AI generated art suffers from an inherent creative limitation. It’s a derivative and subtractive tool. It can only remix what already exists.

There's little evidence that this is fundamentally different from how our own minds work. We are influenced by our environment, and experiences. The art we create is a product of our material conditions. If you look at art from different eras you can clearly see that it's grounded in the material reality people live in. Furthermore, an artist can train the AI on their own style, as the video linked in the article shows with a concrete use case. That allows the artists to automate the mechanical work of producing the style they've come up with.

It lacks intention and human experience that make art meaningful.

That's what makes it a tool. A paintbrush or an app like Krita also lacks intention. It's the human using the tool that has the idea that they want to convey, and they use the tool to do that. We see this already happening a lot with memes being generated using AI tools. A few examples here. It's a case of people coming up with ideas and then using AI to visualize them so they can share them with others.

This is why a crude MS Paint drawing or a hastily made meme can resonate more than a “flawless” AI generated piece.

If we're just talking about pressing a button and getting an image sure. However, the actual tools like ComfyUI have complex workflows where the artist has a lot of direction over every detail that's being generated. Personally, I don't see how it's fundamentally different from using a 3D modelling tool like Blender or a movie director guiding actors in execution of the script.

I can see some applications of AI generation for the more mundane aspects of creation, like the actions panel in Photoshop.

Right, I think that's how these tools will be used professionally. However, there are also plenty of people who aren't professionals, and don't have artistic talent. These people now have a tool to flesh out an idea in their heads which they wouldn't have been able to do previously. I see this as a net positive. The examples above show how this can be a powerful tool for agitation, satire, and political commentary.

Those tools still require direct engagement with the creative process

So do tools like ComfyUI, if you look at the workflow, it very much resembles these tools.

the argument that the photographer composes a shot and manipulating light. In contrast to AI which automates the creative act itself

I do photography and I disagree here. The photographer looks at the scene, they do not create the scene themselves. The skill of the photographer is in noticing interesting patterns of light, objects, and composition in the scene that are aesthetically appealing. It's the skill of being able to curate visually interesting imagery. Similarly, what the AI does is generate the scene, and what the human does is curate the content that's generated based on their aesthetic.

AI doesn’t teach someone to draw, operate a camera, paint, reiterate, conceptualize, and develop artistic judgment. It lets them skip those steps entirely resulting in outputs that are aesthetically polished and creatively hollow. True democratization would mean empowering people to create.

Again, AI is a tool and it doesn't magically remove the need for people to develop an aesthetic, to learn about lighting, composition, and so on. However, you're also mixing in mechanical skills like operating the camera which have little to do with actual art. These tools very much do empower people to create, but to create something interesting still takes skill.