It would increase the number of bad descriptions and desincentivize people to add good descriptions or (depending on the interface/permissions; you are assuming I could just manually go and fix someone else's alt-info in their image) submit fixes and corrections. And it would kill the environment even more anyway.

wouldn't using standard ocr be easier?

Sure. I don't care enough to argue. I'm not going to be the one who implements this system anyway, lol.

i used tesseract and then proofread it for text images(???????)

people do them lots and lots on parts of the fediverse that are not lemmy! in fact, i think the only reason no lemmers do it is because the ui sucks (the alt text box in a post pops up in a completely unrelated section when you add an image!! alt text must be attached to images)

people think local ai is the panacea, when ai must have a shit-ton of content scraped from the internet, and ~countless hours churning in datacenters, for the model to be produced in the first place

ackshyually ai is a broad term that applied to ocr when it was first made, but nobody uses it that way anymore, they refer to machine learning. but wait! machine learning is used for ocr!!

!!!

it's not a sudden, unexplained moral panic, it's a bubble with billions being invested in for no return, and ai chows down on those billions well, thus giant, unexplainable, strikingly accurate ai, that investors want put in every facet of your life. and no one! likes it!!

LLMs can't see. are you talking about one with a vision thingamabob bolted on? you could just use ocr