The case for AI in #accessibility is a hotly contested one but I do feel the baby's being tossed out with the bathwater just a tiny bit. Yes, it is bad that AI is being used to phase out hoomans in all sorts of pursuits. And yes, it is also true that at least at the moment, Ai-generated anything is generally lower quality than hooman-generated stuff. And yes, it is also true that we're seeing AI in places we really shouldn't be seeing it (MDN anyone?) and that people, just like always with a new toy, are going absolutely nuts with it and putting it front and center like it's Cthulhu's new miracle to end all toilet paper shortages. But it CAN, at times, actually be an enabler. It CAN, OCCASIONALLY, actually be used for good, and I don't think people who find this out and do this should be villified
If image generation gets good enough to make me Twitch banners, or blog featured images, or make small video edits for me, or write me CSS that actually looks good, you bet your glutal hemispheres I'm going to make use of it. Purely a matter of efficiency. We haven't had a decent video editor that's #accessible as long as I've been alive, I think i've waited long enough for hoomans to step up. If AI can do it, AI can do it. Screen readers can't tell me if CSS is doing what I want, so if AI can do it that just means I can be more productive as a developer. Again, I've waited for this situation to improve in #screenreaders for my entire professional career, so if what is for all intents and purposes a new tool can enable me to do this, I will happily make use of it. Summarizing articles in between all the newsletter popups, ads, images without alt text and silent videos? Absolutely. Just another #accessibility tool at this point. So yeah. Bad, but not all bad.
@zersiax@Binder It is possible to use AI ethically, but that seems to be more the exception than the rule. Are you using an image generator that is properly paying for images, or one that scrapes and steals them? If someone isn't sure of the answer, they aren't using it ethically. In which case, I'd say they do deserve vilification for contributing to the problem. No amount of "but this is easier for me" makes that okay.
@jjLitke@Binder there absolutely is something to that. Then again, to what degree is an image being " stolen" if it's used as an inspiration for a new creation? What I know of LLMs in general is that really all they do is make plausible predictions, based on what they've seen before. So if images get scraped, provided the dataset is large enough, wouldn't that mean that somebody's ill-gotten art is only used as inspiration for, maybe, a tenth or even a hundredth of the newly generated piece of content? I might be completely beside the mark here, I'll be the first to admit that. If people's images are being literally reproduced one to one that is not OK , particularly without attribution, but is that always the case?
@zersiax The thing I find a lot of people, especially on here, don't seem to understand is that it will improve. In the next five years we're not gonna have the same tools that we do now, and while I'm definitely sick of the hype I don't think AI/ML is inherently bad. There are a lot of really good use cases.
@ZBennoui I mean ... there is the fact that it takes ungodly amounts of resources to run which isn't great for the planet, so in that sense this isn't a great future-proof strategy :)
@zersiax my concern is that it’s going to make people complacent and they’re going to stop fixing things and claiming that the AI can do it for them so why should they bother.
@evilcookies98 shrug. People will find a way to do that regardless. Either they'll eat the cost of the ADA slap on the wrist, we see this a lot in eCommerce, or they'll just flat out not care. Look at companies like Notion, which is at this point almost ubiquitous in the notetaking space, and I just can't see this development as all that much of a game changer. Sure, its another excuse, except this one it might actually be true that AI can get us out of a human-made shit heap. Is that a great thought? Absolutely not. But take it from someone who works in the accessibility field: If it isn't AI, it'll be something else
@zersiax The case I like to make is that neural networks aren't new technology, what's got Silicon Valley excited is making them larger than ever before. (personally I don't find this strategy interesting, & I question its value)
We've long used it for accessibility! We should continue! But then its banal.
The narrative & the reality of this tech seem to be frustratingly different at the moment... Its the narrative that's harmful, not so much the tech...
Add comment