Please don’t rely on automated tools to write alt text!
Sure, AI is getting better at being able to recognise what’s in an image, but its context that matters!
See my example, where the same image, in two places, needs different alt text: https://www.craigabbott.co.uk/blog/how-to-write-good-alt-text-for-screen-readers/#forgetting-the-context
#accessibility #a11y
Sure, AI is getting better at being able to recognise what’s in an image, but its context that matters!
See my example, where the same image, in two places, needs different alt text: https://www.craigabbott.co.uk/blog/how-to-write-good-alt-text-for-screen-readers/#forgetting-the-context
#accessibility #a11y
Comments
But actually also wondering if the AI tools could sit on the other side. As by your example
There are examples where it labels offensive things in the image which aren’t there, like mistaking which finger is up or which way around a hand is for example.
When writing alt text there’s a fine balance between providing enough context and writing too much unnecessary noise.
But let's not overthink - it's just an example.
I think it highlights the nuance of the original issue and how important context is. If it’s difficult for us, it’s even more difficult for an AI.
It also highlights the importance of user research and usability testing. That’s what should drive these decisions.
For example:
Alt: A tabby cat wearing a bow tie, sat in front of a door marked with the number 10.
Caption: Larry has been Chief Mouser to the Cabinet Office at 10 downing street since 2011.
Very thankful for your article which is truly excellent, adding it to my vault for when I teach