Friends are a pretty big deal on Facebook — the social network is always highlighting friendly relationships with cute videos, and everyone understands the implications of being unfriended. However, during Wednesday’s quarterly earnings call, CEO Mark Zuckerberg revealed that he wants the social network to develop an artificial intelligence that can bypass friends entirely and place content that it thinks you’ll like right in your News Feed, regardless if anyone you’re connected with has ever liked it.
Zuckerberg talked about artificial intelligence briefly in a Q&A at the end of the call after first revealing that Facebook made a tidy profit of $3.57 billion that quarter. When asked how A.I. would factor into Facebook’s plans going forward, Zuckerberg said there were two main areas he was looking at. The first was an A.I. that could identify visual content (a photo or video, for instance), understand what it was seeing, and then decide if you as a user would be interested in it.
“And you can imagine that today, we consider putting things in your News Feed that you’re connected to in some way, right, that are from a friend or from a page that you are following or that one of your friends liked,” Zuckerberg said, explaining how Facebook’s News Feed currently works. “But there’s no reason that we shouldn’t be able to match you up with any of the millions of pieces of content that might be interesting that get shared on Facebook every day, except for the fact that we don’t have the A.I. technology to know what those are about and that they match your interests today.”
Zuckerberg wasn’t exactly clear what this A.I. would look like in action — possibly in part because the technology doesn’t even exist yet. Inverse reached out to Facebook for clarification, but based on Zuckerberg’s comments, this does seem like it would be a pretty big change to the way News Feed currently works. No timeframe was given for when an A.I. like this might be feasible.
The other area where Zuckerberg says Facebook has big plans for A.I. is content moderation. The CEO admitted that there is some very inappropriate content that gets posted to the social network.
“And it’s a minority of the content, but I’m really focused on making sure that our company gets faster at taking the bad stuff down,” he said. “And we can do better with people, but ultimately, the best thing that we can do is build A.I. systems that can watch a video and understand that it’s going to be problematic and violate the policies of our community and that people aren’t going to want to see it and then just not show it to people before that experience.”
Zuckerberg is probably right that A.I. can do this unsavory job the fastest, but his aside about doing “better with people” is telling. Human moderators on Facebook already have trouble telling if an image is inappropriate or if it’s an iconic photo that demonstrates the horrors of war. It’s hard to imagine A.I. mastering those cultural subtitles and context clues.