Post #1488452

A year since I first got my Meta Glasses, they remain a reliable conversation starter, and a somewhat unreliable accessibility tool. That is, at least, if you consider their main value as being integrated AI. While it has improved somewhat since I first got them, Meta's LLM is well behind the pack, and has been not only unhelpful but also at times actively sassy. Recently it told me an LP was called "Title of Record". After repeated requests to read a document, it eventually said "it's easier to go paperless". Sure, mate. Even so, I have no regrets about buying them, and use them all the time. As a wheelchair user, having a camera on my head, freeing up both hands for wheels and cane, is invaluable. A real, human co-pilot has helped me out of many scrapes which would have otherwise been far more difficult to work through. My glasses have allowed me, for the first time, to film watchable video. I've always been hopeless at getting my intended subject in frame from my phone, but I know whether or not I'm looking at something. When house hunting recently on my brother's behalf, I filmed walkthroughs of each property which reliably caught the details he needed to see. Through better trained AI-based Audio Description apps, I can even independently check what I've captured. It's a shame that Meta AI is so bad, and needlessly handicapped, in Australia at least, by the unavailability of video streaming AI. But every new technology comes alive in the hands of users in ways that weren't anticipated. I credit Meta for recognising, however belatedly, that blind and vision impaired people are part of the audience, and making the glasses a platform that allows other developers to unlock more of their potential. But even if they never improved, the uses I already have for them make them a worthy addition to my personal toolkit. So was the hype real? Did some of us adopt too early? I think the answers are complicated, and I'm interested in what others have found after using the glasses for a while. Where do they work well, where do they not live up to expectations, and what unexpected use cases have you found?