Art (real art) at it's core, is an extension of the artists soul. It is personal. It comes from the unique perspective of another human being. Could a true AI make art? Absolutely, if it could express it's own perspective. Algorithmically generated art is not that, it is the defacing of the personal experience.

the 'AI-art' models don't do it on their own, they necessarily require a unique perspective of another human being or a few and (sometimes more sometimes less) creative trial-and-error constructing and formulating of artistic ideas for the medium through prompts, edits, multiple layered passes, etc. It's not a particularly fulfilling medium or process to me personally, but there is an active undeniable component of human perspective being applied in creative ways. Bad art, lazy derivative art, and plagiarism has always existed; and has always been recognized as such by the critical observer. Same with direct forgeries. The means of them being made easier to create and being socially and societally problematic for the climate or the job market doesn't change what the thing "is." Nor does the tools used becoming "smarter."

I've been a studied-and-practiced artist most of my life in many mediums and forms, traditional and digital, visual and musical and textual and spoken, you name it. The term of "art" is not as restrictive as you think it is. And speaking about "real art" smacks of a self-importance; because there is no definition of "real" vs "fake" art that you can create that doesn't dissipate when you zoom in on any of its delineations. A banana taped to a wall is "real art" in the context of its creation and purpose. The artist didn't make the tape; they didn't build the building and erect the wall --- nor finish and paint the wall to which it is taped. They also didn't grow the banana or transport it to the store to be bought. What makes this different, materially, not just how you personally "feel" about what it 'does to "real art?" Is photography art? They're often only capturing what is already there; and a lot of photography these days is done digitally with all sorts of AI focus, recognition, and lighting tools. Is someone using AI face-recognition fuzzy select tool to crop something in a digital artwork or photo edit 'no longer making art' because they didn't do the work themselves? If they use a machine-learning tweening process in an animation? What if they print an AI-art image and then do hand-work on it, or use components of it in another piece? You run into a Ship-of-Theseus problem --- when does it stop and start being "art?" At what point does "not real art" become "real art" and vice versa?

I make art and I also have known a lot of artists over the years. imo artists in general have an over-inflated and even mythologized sense of themselves and what they do, and I've felt they demonstrate it when they fight to distinguish some core spiritualistic difference between what they do and what others do (often doing petty-bourgeois cudgeling of other laborers). And with these machine learning models too, people are using tools which are already among the countless "AI" tools of more-or-less complexity already widespread used in digital painting, animation, modeling, music production, in both digital photography and photo editing, in both digital videography and video editing, and even to some degree in writing (predictive texts, spellcheck, formatting suggestions,) etc; and even these image models tools themselves at their current stage "learn" really not all that much different from people on a fundamental level, as any artist who has seriously studied and understands how they work could tell you if they're being honest with themselves.

Artists don't try to depict an image from what they "know," they do repetition after repetition of what they actually see in reference (including from other people, and from other artworks copying their favorite artists over and over regardless of what that artist wants) drawing humans and objects and representations of them over and over in different angles, and lightings, and stances, styles, and in their individual parts and aspects (including angle, shapes, planes, etc.) over and over and over etc. to train themselves to break down and be able to reproduce and utilize these things and aspects of things in their different variables, angles, forms in different contexts, to be able to reproduce and utilize these objects or aspects when and where they are called for ie. "prompted" (as an internally-driven prompt or an external commission request) from the agglomerated experience. People who draw what they "know" are going to make things worse than even the wonkiest image model lol. That's the first thing you learn is to not draw what you "know," draw what you see. Nobody learned how to do perspective drawing by drawing what we think proper perspective looks like. We had to break it down into formulaic and often binary constructs and rules and then do mass-repetition for various objects in various perspectives to begin to slowly make them more and more after endless repetition align with the actual proper perspective. The biggest difference between us and these particular tools is just that we have the ability for fluidly abstracting conceptions-of-objects external of their visual presence; where a machine learning image model doesn't; though even that can be approximated with tagging systems and metadata on the training data. Again, not particularly fulfilling for me personally, and there are other (much more material and important) implications and issues with its current applications and ownership and consumption in our current societal organization; but that doesn't change what it is or what it does or the relationship to art.

Artists complaining about this as some soul-doom insult to "real art" are, in my opinion as an artist, engaging in petty-bourgeois self-obsession; because automation has been destroying working class relations to their labor for like a decade and a half causing all the same disgust and panic and collapse of self-sense and self-worth and philosophical dread and alienation as well as removing peoples material wages they need to live for 15-20 years. And people have been organizing and rallying around fighting it or working at challenging its material and social implications and the question of its ownership; but petty bourgeois artists as intelligentsia didn't care because it was only the "unskilled blue collar peons unlike my called-from-high spark of divinity ~craft~."

I can't tell you how many times I've heard from artists of all kinds "yeah machines can do A B and C jobs but they'll never be able to xyz" But now artists are facing a material and philosophical undermining on many fronts just like all laborers have been from automation for 15-20 years and maybe once the ego-shattering ends they can realize they have to throw in with the workers as workers and stop over-inflating their importance. I'm an artist, but this stuff is to me not much different than any worker watching a machine do any job they dedicated their life to perfecting, but it does it in a fraction of the time and gets better and better at it but without any of the "soul" (which begs the question, what is the "soul," in what I do, and was there one? hence the spiraling philosophical horror). In fact, this exact phenomenon is a tale as old as John Henry.