zlacker

>>dredmo+(OP)
I've been finding that the strangest part of discussions around art AI among technical people is the complete lack of identification or empathy: it seems to me that most computer programmers should be just as afraid as artists, in the face of technology like this!!! I am a failed artist (read, I studied painting in school and tried to make a go at being a commercial artist in animation and couldn't make the cut), and so I decided to do something easier and became a computer programmer, working for FAANG and other large companies and making absurd (to me!!) amounts of cash. In my humble estimation, making art is vastly more difficult than the huge majority of computer programming that is done. Art AI is terrifying if you want to make art for a living- and, if AI is able to do these astonishingly difficult things, why shouldn't it, with some finagling, also be able to do the dumb, simple things most programmers do for their jobs?

The lack of empathy is incredibly depressing...

>>meebob+kc
I have crossed over the other direction from coding to drawing and suspect that neither side understands their craft well enough to assess what'll happen.

Most of coding is routine patterns that are only perceived as complex because of the presence of other coders and the need to "talk" with them, which creates a need for reference materials(common protocols, documentation, etc.)

Likewise, most of painting is routine patterns complicated by a mix of human intent(what's actually communicated) and the need for reference materials to make the image representational.

Advancements in Western painting between the Renaissance and the invention of photography track with developments in optics; the Hockney-Falco thesis is the "strong" version of this, asserting that specific elements in historical paintings had to have come through the use of optical projections, not through the artist's eyes. A weaker form of this would say that the optics were tools for study and development of the artist's eye, but not always the go-to tool, especially not early on when their quality was not good.

Coding has been around for a much shorter time, but mostly operates on the assumptions of bureaucracy: that which is information is information that can be modelled, sorted, searched. And the need for more code exists relative to having more categories of modelled data.

Art already faced its first crisis of purpose with the combination of photography and mass reproduction. Photos produced a high level of realism, and as it became cheaper to copy and print them, the artist moved from a necessary role towards a specialist one - an "illustrator" or "fine artist".

What an AI can do - given appropriate training, prompt interfaces and supplementary ability to test and validate its output - is produce a routine result in a fraction of the time. And this means that it can sidestep the bureaucratic mode entirely in many circumstances and be instructed "more of this, less of that" - which produces features like spam filters and engagement-based algorithms, but also means that entire protocols are reduced to output data if the AI is a sufficiently good compiler; if you can tell the AI what you want the layout to look like and it produces the necessary CSS, then CSS is more of a commodity. You can just draw a thing, possibly add some tagging structure, and use that as the compiler's input. Visual coding.

But that makes the role a specialized one; nobody needs a "code monkey" for such a task, they need a graphic designer...which is an arts job.

That is, the counterpoint to "structured, symbolic prompts generating visual data" is "visual prompts generating structured, symbolic data". ML can be structured in either direction, it just takes thoughtful engineering. And if the result is a slightly glitchy web site, it's an acceptable tradeoff.

Either way, we've got a pile of old careers on their way out and new careers replacing them.