Firefox 130 will feature an on-device AI model that automatically generates alt-text for images, integrated into its built-in PDF editor.
our end goal is to make it available in general browsing for users with screen readers.
So obviously the people working on it know that the appropriate place for this stuff is in the screen reader, and/or in software that is designed for producing web pages. Who made the decision to cram it into the Firefox PDF editor instead, and why? Is anyone actually using that when they want to create a new PDF document? Is it just for PR reasons so they can claim there’s AI in Firefox now? To impress us, or to make way for more of the same? What is actually going on Mozilla?
But anyway it’s a difficult problem, as the example image suggests. It’ll be interesting to see how much progress they make.
I don’t know why, but I feel like putting the new experimental feature in your niche use first, possibly where alt text barely exists with how people generate pdfs, so getting feedback is to a specific audience who need it most while impacting nearly everyone else, does seem like a logical first step.
I got to compare that to Google, who put it front and centre of their namesake product.
