Losing a single earbud in a gravel field is usually a write-off. One person turned it into a demonstration of practical AI vision: they photographed the ground and asked ChatGPT to spot the missing Bose earbud in the images — and it worked, TechRadar reports.

The approach is clever precisely because it plays to what multimodal models do well. Picking a small dark object out of visually noisy gravel is a needle-in-haystack task where, as TechRadar puts it, the chatbot proved to have better eyes than humans. Systematically photographing an area and letting a model scan each frame beats crawling around squinting at rocks.

TechRadar concedes it's probably not the most environmentally friendly earbud-finding method — burning inference compute to locate a $100 accessory has a certain absurdity to it — but it beats the alternative of buying a replacement pair.

It also hints at a broader pattern: as vision models improve, 'find the thing in this photo' is quietly becoming one of their most genuinely useful consumer applications.