Maybe we can start using "machine learning" for legit and useful applications and "AI" for bullshit that's only use is to steal people's work in order to replace them. Since they've already done most of the work for us.
Instead of baking a contextless AI-generated description into the alt text, wouldn't it be better to keep encouraging people to write proper descriptions of the image and instead build this into screenreaders to serve as an optional last resort for images without alt text?
Woah! This’ll help for Bluesky for sure!
I never really added alt text to images in my posts on Twitter before I came to bluesky, people here wanted me to so I did, but that’s gonna be cool!
Wonder how it’ll play out on lower end devices? Assuming it’s not just OCR recognition 🤔
Really? Instead of you speaking to people who need alt text, you'll just get the computer to do it (poorly). They can interact with the computer instead of with people? Wow. This is huge, but in an unsurprisingly AI shit way.
It’s not a fox made of fire encircling a globe? Do you see the issue I’m driving at? The screencapped example description lacks texture and depth. Not everyone would know what it is merely by tagging it as Firefox’s logo, especially since it used to be a phoenix. This is a small issue here, sure,
I'm curious if it'll add Alt text to images that don't need it, like element borders and other layout centric images, or blank pixels used for copy protection
Not actually useful from an accessibility perspective. Possible detrimental because people will rely more on this tool instead of writing good alt text.