You might recall a couple of years ago when the “mannequin challenge” was making its rounds on social media. For those who are unfamiliar, it is where a group of people in a room hold weird/funny poses while the person with the camera walks around, making it seem like everyone is a mannequin in a store.
While those challenge videos were fun, it seems that they are now being used to help train AI. That’s what Google is doing where according to reports, Google is using all those mannequin challenge videos to help train its AI to better predict depth. This is because unlike human perception, AI still isn’t quite there yet in terms of being able to figure out depth perception during motion.
So by using these mannequin challenge videos, which shows people staying still (as much as possible) while the camera pans around, it’s actually the perfect data that could help train an AI’s depth perception in motion. While it seems that the AI isn’t used for nefarious purposes, the fact that they are used without the people in the videos being notified has raised some privacy concerns, although whether or not you can do anything about it is unclear.
However, it does make you want to think twice the next time you decide to publicly take part in these challenge videos in the future.