i've been reading a lot of web novels lately and it's kind of disappointing to see how pervasive the ai art use has become. you'd think as fellow creatives they'd "get it" more.
otoh i don't really see ai text gen as being as problematic as ai image gen (even though i probably should), so maybe it's kind of the same, where i'm close enough to it to see it as worse, but for someone who isn't a visual artist, they aren't.
@xvln I see AI generated anything in there, I dump the piece and block the "author".
If you need the computer to tell you what to write/draw, you're not a creator.
@xvln that seems to be quite often the case indeed: for “our” art it's easier to understand/feel why genAI is bad, for other forms it's more difficult (one basically has to go by analogy).

@oblomov @xvln I dunno... from my point of view (as a non-artist, non-writer, but lifelong writer of software), "AI" slop is slop pretty much regardless of the medium.

It *really* stands out for me in art and the written word, where the quirks and limitations of these tools are pretty obvious and distracting.

I think it can be harder to see in software if you're not looking at the source code, because the slop is often buried a bit deeper below the surface, and the tells are more indirect.

@swetland @oblomov my dayjob is as a software engineer at a company where use of AI for code generation is non-optional, sadly. but as a result i kind of... am not as bothered by AI code generation. i think it's not at the quality of human written code, and that's a problem, but the principle of the thing doesn't bother me the way it does for AI visual art. not sure if that's a good thing though. probably not.

@xvln @oblomov The principle of the thing absolutely bothers me. I've been writing software for 40 years, ~30 professionally, and I find the dependence on these tools appalling and incredibly depressing.

I'm fortunate to be in a place where I don't have some asshole managers telling me I have to use these tools instead of doing it right. Sad to see the industry destroy itself.

I enjoy designing, building, and *understanding* systems top to bottom and "ai" tooling is antithetical to that.

@xvln @oblomov I suspect it's possible to use some of this stuff in moderation, but I really have no interest in ceding my thinking and comprehension to tools that do not actually think or comprehend or have any ability to know what is real or correct.

Hell, half the fun of the work I do is constantly learning new things and I certainly have no desire to give that up.

I also, maybe arrogantly, take pride in my work and believe there is art as well as science to good systems design.

</rant>

@xvln best use case for llm text from everything I observed is propoganda. It's looks natural and as if you having conversation, but really it just use your irl time to argue with sophisticated bot. Scammer that orders your art then to lead you to send money first to check if your bank legitimate. Politics where intent to either convert or waste your time on telling why argument is wrong (also extra time to check your sources and provide them as well). Sell offers on AliExpress and other cheap market where they will use LLM to have believable translation and sell you LLM art that you only see is wrong after receiving and paying attention (that time when you wish they stole it instead... Which they done but smear it in shit on top).
So LLM is useful if you know what you doing and only check for errors... But that's not how LLM used right? It's a slap on feature and heaven for scammers to destroy and undemine anonymity be it soft,art, orders online, foss, online debates and many more.

There whole ecological and industrial impact (price on GPU/hard drives hey ho!).

People build houses using asbestos, well here your digital equivalent of that. 

@xvln i consider both to be highly evil
both for destroying the planet and for lowering standards and making people even dumber