Mozilla really needs to reconsider this whole AI generated alt text endeavor.
-
Mozilla really needs to reconsider this whole AI generated alt text endeavor.
-
beaware@social.beaware.livereplied to cameron@route66.social on last edited by
@cameron @Adventurer they shouldve just made a deal with OpenAI. I use ChatGPT for most of my alt text and it does amazingly.
However, they don't care enough about accessibility, so I doubt this will improve *at all*.️
-
hugegameartgd@mastodon.gamedev.placereplied to beaware@social.beaware.live last edited by
@BeAware @cameron @Adventurer OpenAI content is also AI-generated though
-
beaware@social.beaware.livereplied to hugegameartgd@mastodon.gamedev.place last edited by
@HugeGameArtGD @cameron @Adventurer
So...?
Is your stance, "accessibility, as long as it's not AI"?
If so, that's a strange stance to take. I can't wrap my head around how to accurately describe an image.
So it's either that, or visually impaired folks don't know what my image contains and I'd much rather have better accessibility using AI than none at all....
-
frumble@chaos.socialreplied to beaware@social.beaware.live last edited by
@BeAware Are you really suggesting they should have rather chosen a proprietary cloud provider instead of a local model (which will only getting better from here on)?
-
beaware@social.beaware.livereplied to frumble@chaos.social last edited by
@frumble I'm suggesting that Mozilla might should have made a deal with OpenAI.
I just didn't think Mozilla would offer any local models to do this sort of thing and was suggesting that if they *must* do AI alt text, surely there's better options than the example pictured above.
I don't think people would want to download a 5-10 gb model to make a paragraph or so to describe images, though maybe I'm wrong.️