Christine MalecFor the first time in my life on Masto, I read alt text that made me wish the person hadn’t bothered. After a wordy post, they wrote in the alt text field something like, “I don’t really want to write alt text and train the AI for free, but okay, person holding a computer.” I’m going to assume they didn’t think about what a slap in the face this was to me, and possibly to anyone else who relies on alt text. 3 Times I started an acerbic reply and deleted it, but I’m still seething, so I’m asking anyone who is so disingenuous to just not bother; you’ll create less hard feelings that way.
Climate Jenny 2.1@ChristineMalec That’s weird. Why do they think it has anything to do with training AI? Maybe I’m just ignorant, but I thought alt text was for humans.
@ClimateJenny No idea, goodness knows the body of their post had enough verbage in it.
AaronThe reason is that alt text is scraped in order to train multimodal text + image models.
But for anyone who actually listens to the blind community, it doesn't take long to learn that these models are pretty important to a lot of people for accessibility. And regardless, alt text is directly helpful. I'm happy to put my displeasure with big tech on hold for things that genuinely improve people's lives.
Dilman Dila@hosford42 @ChristineMalec @ClimateJenny In that case, I think writing subjective alt text would be a solution for those who fear it trains AI. Something like, "I took this pretty flower near sunset, there's a bird in the background spoiling my perfect view of the sunset and the flower is..." I like writing my alt text like that, not thinking of AI but because I suck at describing images accurately and so put in my feelings, which I suspect would be worthless to AI?
C++ Wage Slave@dilmandila @hosford42 @ChristineMalec @ClimateJenny
Another approach would be to write clear, descriptive alt text but use something like Nightshade to poison your image so that "AI" can't make sense of it:
I still think that's working against the public good. If the models being trained on this data are used for accessibility by the blind community, and we are intentionally working to lower their accuracy, we are hurting the blind community. We should fight the AI vendors on a front where there aren't innocent bystanders who will be hit.
Author-ized L.J.@hosford42 @dilmandila @ChristineMalec @ClimateJenny I think calling image poisoning harmful to accessibility is a bridge too far, though. Should artists hand over their online art for free to tech companies for the sake of accessibility? Surely there's a way to aid accessibility without giving everything to scrapers indiscriminately. (For clarity I'm not defending alt text tampering, which I agree is a jerk move.)
Sure, that's poisoning the image itself, not for the purpose of ruining models that could generate alt text, but ruining models that could generate othe images of the same style. I say go for it.
@hosford42 Is there a way to do one but not the other, though? Do alt text-generating models and generative diffusion models use different datasets or get their data differently? 🤔 Ideally there would be NO non-consensual scraping of any websites, and models that purport to have pro-social uses shouldn't be resorting to it.
I think the core problem here is that the laws need to catch up to the reality -- and be enforced, too. Someday we will live in a world where we each have irrevocable ownership of our own data, and can choose to donate it all or in part to those causes we deem just. I have no idea how long it will take us to get there, though. Probably not our lifetimes, from the look of it.