The internet has become 3 massive multi-user blogs, each largely consisting of screenshots taken of the other two. This kind of blows, and not just for the usual reasons that may spring to mind.
Images are a terrible medium for online communication! Not everyone online uses a monitor. Any messages contained in a picture is straight up unacceptable without alt-text. It also makes it harder to find and fact check sources, or to spread a thought or idea further than yet another image upload. Copy/pasting text is just plain easier than downloading and uploading.
If you're going through the trouble of creating an image post, take an extra minuite to copy/past (or even transcribe) the source text into the alt-text submission. It's not much, but it goes a ways to improving how we use this blasted network!
https://uxdesign.cc/how-to-write-an-image-description-2f30d3bf5546
100% agree on this. This also strikes me as something that AI would actually be good at. I’d be shocked if there aren’t screen readers or other software that have an LLM built in to describe images and answer questions about them or software at least working towards that. Obviously it may not always be perfect and bespoke alt text is going to be better since the LLM might not get the gist of certain memes etc but yeah seems like a tailor made use case as opposed to the current shoehorning of AI into nearly everything without any real thought so that half the internet becomes pure slop..
I think one of the core problems of using AI is that it doesn't know why the image is there, so even when talking about a relatively straightforward image (i.e. not some meme which requires a bunch of background knowledge) I have serious doubts that it would focus on the pertinent details even if we assume it makes no actual mistakes. Obviously, it's nice for people to have something they can use when alt text isn't present, but it should be a backup rather than the first line option.