So I've been thinking about why is it that most programmers I know are excited about AI replacing a lot of what they do and most artists I know are not. What's especially odd is that as far as I can tell, coding is a much more solvable problem because of huge data sets and high checkability. Ideas:
Comments
Programmers: "I copied this code from some forum that predates the dinosaurs and somehow, SOMEHOW, it actually works. Oh, there's a new tool that does my job at the press of a button? AND I can blame it if it's code breaks? AI ROCKS!"
...you just hit this thing with a hammer, right?
Yet Another High-Level Language: communicating precise requirements to an AI can be seen as just another step in the evolution of programming languages.
Developers who cannot even reach that quality themselves are either impressed or scared because it outperforms them.
Using it for anything other than simple writing is a waste of time and resources.
The no-fun part of coding. So an AI to slurp that & answer how to *use* the API is a lifesaver.
2) Economics - more programmers expect to still be useful and well-paid even if mostly replaced, while artists expect to get kicked to the curb
4) Grind - in art, grind is considered part of the process, sometimes even meditative; in coding it's an annoyance
7) Ingroup/outgroup stuff - artists reasonably fear that they are not *in* with powerful tech players the way tech geeks might. Tech geeks see AI people as *us*
In contrast, art is personal expression so an amalgamation of a data set with their work as a training model feels...weird
10) Something else?
(Which might make sense, as I *also* draw, and, I'm married to an artist.)
I think the people that most benefit from genAI are in tech, or working on pipelines, or new to it. And I assume excitement tracks
I think the people's work that was used for training were long standing open source developers.
I question how much overlap there is
I don't actually *enjoy* the coding part, I enjoy solving a problem / creating a product.
I mean really, libraries already do basically this same thing; someone else has done a lot of the coding and you just stitch some stuff together.
Aesthetics is a completely separate endeavor from engineering! One is focused on direct intuitive satisfaction, and the other on indirect material means for the same
Open source complicates this but we gave that away to _people_, not so corporations could profit from it!
However it is doing real harm, & the little good it's doing doesn't balance the scales
I've seen friends use it to learn how to use an API. Mostly because documentation for coding is mostly awful!
For actual coding though it's pretty bad. Banned at work due to copyright 🤷♂️
As a result the only time I ever see anyone using AI is for the busywork no one wants to do.
I just wrote a test with it, in less than 1/4th the time it would have taken to do it by hand.
In a few years you won’t make it past an interview if you can’t find the use for AI.
Also I'm not saying there aren't uses for LLMs, but the bullshit vs reality ratio is heavily towards the bullshit.
Personally I dislike my motivation being "scramble to live or die penniless" 🤷♂️
While some code can be masterfully written, it has no value on itself, beyond the features or performance it builds.
You learn to not get attached to your code and keep replacing it with better code.
Art is not like that.
Art is not like that.
The difference between "our programmers get more done" and "we no longer need artists"
So they can't whine too much when automation comes for them.
AI will not be meaningfully different
fundamentally: software is made of decisions (https://siderea.dreamwidth.org/1219758.html), and *someone* (not a machine) needs to be accountable for the decisions made in the software
Skills getting obsoleted on the regular is just how coders live.
it feels like every group project has one person who pastes in hundreds of lines of GPT garbage that doesn't even compile and gets mad when i remove it
Maybe brand/IP is a bigger problem for art
The HARD part of programming is design and abstraction, and I worry that making programming more accessible is going to lead to problems
In other words, programmers are probably a bit spoiled by the good times. If AI came a few years later, maybe we'd feel differently.
If you can grab code off stack exchange, or a library off GitHub, or have a LLM spit it out and it works, use it.
But I’d caution against using power-hungry gen AI if this problem is already answered, and watch out for hallucinations! Test that code.