Right now there is a bit left to be desired when it comes to lemmys accessibility features, but it’s a good idea to be mindful of the fact the fediverse and its platforms tends to have pretty universal accessibility features that will likely come to lemmy sooner rather than later
Or somebody should build AI-powered image interpretation into screen readers.
Unfortunately those aren’t very reliable, as our current iteration of AI is not very reliable. Most models to use from perpetuate a multitude of different bias pretty heavily.
And also, don’t they need quite the processing power to be useful? Anybody know?