The company announced a new project on March 15, called the Learning from Videos, which is meant to improve the company’s “core systems” and “power new applications”. The company said it would help improve both its moderation efforts and recommendation engines for 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 recognise the nuances and visual cues across different cultures and regions. And by helping AI researchers break away from the reliance on labelled data, we can improve AI-powered products and create entirely new experiences,” the company said in a blog post.
The timing for the project is curious and seems to be an effort at keeping Facebook’s AI systems up to date. Many experts have said that interest-based algorithms developed by platforms like TikTok are a threat to how Facebook’s algorithms operate. Even Google recently said it would be moving to an interest-based system for advertising, after it starts blocking third party cookies on Chrome.
Case in point, one of the first applications Facebook is enabling with this new project, is on Reels, the company’s replica of TikTok. Facebook says it will help improve the videos it recommends to Reels users, an essential component of how TikTok kept its users engaged. In its blog post, the company said the algorithms will learn “themes” from the videos users watch and group similar videos under those themes. The algorithms are designed to learn audio, visual and textual input in videos.
It’s unclear whether Facebook plans to use these for ad targeting right now, but the company did tell The Verge that it won’t seek special permission from users to use their videos as training data for AI. The company’s data policy informs users that uploaded content can be used by Facebook for research and development. – Mint, New Delhi/Tribune News Service
Source: Tech Daily Report (techdailyreport.net)