3 ms·

> I strongly disagree. Empty alt text means no image description. Think about that. You are blind. Your Screen reader tells you "image", nothing more. I feel l

> I strongly disagree. Empty alt text means no image description. Think about that. You are blind. Your Screen reader tells you "image", nothing more. I feel like this is something where AI can actually shine: generating descriptions of images and video for blind people. Best of all it's already a solved problem (and others have thought about it already): https://www.blog.google/outreach-initiatives/accessibility/get-image-descriptions/ https://www.blog.google/outreach-initiatives/accessibility/g... The next step is using AI to describe (important) changes to pages. Think of a chat on a website with new messages appearing. A voice saying: "Alice writes 'hello'" goes a long way. Thinking of it there really ought to be a way for JavaScript to interact with screen readers, allowing developers to give changes performed by JS something like an alt text. The final step would be using AI to make interacting with sites easier. Finding the right button/element to click/focus can be really annoying with a screen reader. If you could just tell the AI "fill the form with X and Y and click submit", or "go to the next page", and possibly having that work even if what you're looking is buried in some context menu, pretty much any non-accessible page becomes accessible.

JimDabell 6y ago
> I feel like this is something where AI can actually shine: generating descriptions of images and video for blind people. That's exactly what you don't want in the majority of cases. The alt attribute is for an alternative to the image, not a description of the image. For decorative images, the alternative is the empty string, no matter what the image contains. "Laughing group of people" might be a correct description of a stock image, but completely wrong as alt text. For, e.g. a red arrow pointing out a required field, the alternative is "Required", not "Red arrow pointing right".
goldcd 6y ago
I think it might already be there - I just right clicked in Chrome to see what options I could pull up for an image, and noticed a "Get image descriptions from Google" option
armoredkitten 6y ago
I wouldn't consider generating descriptions a "solved problem". There has been progress with AI, but part of the issue is that image classifiers are trained to describe the objects in the image, but not necessarily the feeling or meaning of the image. If you have Instagram, scroll through the alt text on the images some time. You'll see descriptions like "May contain: Person standing outside". That may be an accurate description of the picture, but it may be a poor substitute for what a sighted person sees -- someone showing off their new sweater, or posing in front of an interesting wall, or an aesthetic splash of colours. Of course, maybe someone who is fully blind doesn't care too much about a splash of colours, but the point is that describing an image is more than just about identifying the individual objects visible in the image. Because when we see a picture of an adorable puppy, we get more out of that image than just "dog sitting on floor". As of now, human-generated alt text is still the best solution in most cases, even if we are definitely making progress with AI. It's hard enough for humans to describe the aesthetics of an image, let alone a machine.

JimDabell 6y ago

> I feel like this is something where AI can actually shine: generating descriptions of images and video for blind people. That's exactly what you don't want in the majority of cases. The alt attribute is for an alternative to the image, not a description of the image. For decorative images, the alternative is the empty string, no matter what the image contains. "Laughing group of people" might be a correct description of a stock image, but completely wrong as alt text. For, e.g. a red arrow pointing out a required field, the alternative is "Required", not "Red arrow pointing right".

goldcd 6y ago

I think it might already be there - I just right clicked in Chrome to see what options I could pull up for an image, and noticed a "Get image descriptions from Google" option

armoredkitten 6y ago

I wouldn't consider generating descriptions a "solved problem". There has been progress with AI, but part of the issue is that image classifiers are trained to describe the objects in the image, but not necessarily the feeling or meaning of the image. If you have Instagram, scroll through the alt text on the images some time. You'll see descriptions like "May contain: Person standing outside". That may be an accurate description of the picture, but it may be a poor substitute for what a sighted person sees -- someone showing off their new sweater, or posing in front of an interesting wall, or an aesthetic splash of colours. Of course, maybe someone who is fully blind doesn't care too much about a splash of colours, but the point is that describing an image is more than just about identifying the individual objects visible in the image. Because when we see a picture of an adorable puppy, we get more out of that image than just "dog sitting on floor". As of now, human-generated alt text is still the best solution in most cases, even if we are definitely making progress with AI. It's hard enough for humans to describe the aesthetics of an image, let alone a machine.