According to Facebook, the goal is to enable you to search for yourself, or for someone else, through any videos that have been shared with you, similar to how you can currently search through photos of you, not just photos uploaded by you. The feature could also be helpful as a way to find a moment in a video, allowing you to search a point when a particular friend walks in to the video simply by typing in their name, for example.
Another goal for Facebook is to be able to generate captions for what people say in videos, without you having to upload a subtitles file. This could be helpful for the hearing impaired, or simply for when you’re in public and don’t want to play audio out loud.
Of course, all this could have far more serious implications — as artificial intelligence improves and becomes able to recognize people even in a moving setting, it’s conceivable that the same kind of technology could be applied to things like security systems, which could in turn raise privacy concerns.
Facebook is one of a number of tech companies working on advancing deep learning, which is a kind of artificial intelligence that involves training computer-based neural networks through the use of massive amounts of data, then getting the networks to make decisions based on the data it has processed in the past. Snapchat is also working on artificial intelligence for video through Snapchat Research, however Facebook is perhaps further along in its development and obviously has more resources at its disposal.
