@gadgetoid
I think there are two main issues still to be resolved.
This (slightly older) post talks about the reality faced by my friend who is blind when she experimented with it.
https://tink.uk/thoughts-on-screen-readers-and-image-recognition/
Secondly, this post (from a handsome chap with a beard) which talks about the context behind describing an image.
https://shkspr.mobi/blog/2023/07/should-you-embed-alt-text-inside-image-metadata/
So, yes, it *can* be useful. But unless there's a lot of context, it can also be useless.