Developers at Meta have used artificial intelligence (AI) to more accurately select videos that match users' interests, which has contributed to a significant increase in video viewing time on the social network Facebook.
Using AI makes Meta products "smarter"
It is known that before being deployed on Facebook, the AI recommendation algorithm was used in the service Meta Reels, a competitor to TikTok. It would not be surprising if Meta would soon deploy this algorithm to all of the company's products if possible. According to Facebook CEO Tom Alison, since starting to test the new recommendation algorithm on this platform, the time spent watching videos on this social network has increased by 8-10%.
At the Morgan Stanley technology conference in San Francisco this week, Tom Alison shared about the new algorithm, saying: "This shows that the new model architecture learns from data much more effectively than the previous generation. This is a good sign that we are on the right track."
Meta has used different recommendation algorithms on Reels, Groups, and the Facebook Feed. But the success of the AI algorithm in Reels has prompted developers to use it more widely. The move is part of Meta’s larger strategy to integrate AI into all of its products. To prepare for this goal, the company has spent billions of dollars to buy hundreds of thousands of NVIDIA graphics accelerators, which are needed to create large language models.
Source link
Comment (0)