The clearest positive case, and the one least discussed. Real-time captioning, image description for blind users, plain-language rewriting for people with cognitive disabilities, speech interfaces for people who cannot type, and translation for anyone operating in a second language. These are substantial improvements to daily life for a lot of people.
Technical view
Capabilities include automatic captioning and transcription, image and scene description, sign language work at an earlier stage, text simplification, and voice banking and synthesis for people losing speech. Requirements are accuracy sufficient for reliance, latency low enough for conversation, and graceful failure, since a confidently wrong description is worse than none. Accessibility regulations increasingly treat these as expected rather than optional features.
Related
Part of AI Terminology Index, section: Society and Impact.