In what can turn out to be a significant breakthrough if it succeeds, Facebook is trying to teach artificial intelligence (AI) systems to understand what’s happening in videos like a human can. Facebook has announced its latest initiative that is training its AI on Facebook users’ public videos.
One of the biggest hurdles and competitive advantages in AI is access to training data and tech giants like Facebook, Amazon and Google have been collecting resources from their millions of users and have managed to get ahead in various areas. Facebook has already managed to train machine vision models on billions of images from Instagram and is not taking its first steps into making machines understand video.
“By learning from global streams of publicly available videos spanning nearly every country and hundreds of languages, our AI systems will not just improve accuracy but also adapt to our fast moving world and recognize the nuances and visual cues across different cultures and regions,” Facebook explained in a blog.
This project, called Learning from Videos, is a part of Facebook’s “broader efforts toward building machines that learn like humans do”.
Results from these machine learning models will be used to create new content recommendation systems and moderation tools, Facebook said, and can also be used to do more in the future.
If Facebook succeeds with this, AI that can understand content in videos can give the company “unprecedented insight into users’ lives” and allow Facebook to analyse users’ hobbies, interests, preferences of brands and types of clothes etc.
Facebook already has access to information like this via the current ad-targeting operation it already has in place. But, as The Verge points out, “being able to parse video through AI would add an incredibly rich (and invasive) source of data to its stores”.