Facebook To Rely On AI To Crack Down On Fake News, Offensive Live Videos
Artificial intelligence may help Facebook effectively weed out fake news from users' newsfeeds, company executives said, that is if Facebook lays out policies to implement the technology.
Facebook has been the hot target of several publications of late, with generous blame issued on its erring newsfeed algorithm that fails to sort out fake news from real news. The proliferation of fake news on the site, according to several op-eds, was the impetus behind President-elect Donald Trump's victory in the recently concluded U.S. election.
Though Facebook chief Mark Zuckerberg has denied the claims, it doesn't take a genius to infer from the amount of offensive posts and sensitive content on the site that the company sorely needs to update its algorithm to properly send fake news and sensitive down the bin.
AI Will Help Facebook Get Rid Of Fake News
Yann LeCun, Facebook's director for AI research, said that the technology could be implemented to eliminate fake news and at the same time detect violence in live videos by virtue of content filtration on the site, as reported by Wall Street Journal.
The problem is Facebook hasn't quite figured out how to introduce AI responsibly.
During a roundtable discussion in Menlo Park, California with reporters, LeCun gave reporters food for thought about the trade-off between filtration and censorship. The technology, according to him, either exists or can be developed.
"But then the question is how does it make sense to deploy it? And this isn't my department," he said.
Automatic Flagging For Live Videos
Facebook has mostly relied on its users to report sensitive media. But this could change soon. The company could soon flag offensive content in live video automatically, as it works on implementing AI to comb content, said Joaquin Candela, Facebook's director for applied machine learning.
Using AI to flag live video is still at the early stages and comes with a couple of challenges, according to Candela. First, the technology needs to work fast. Second, content must be prioritized the right way so that an expert who knows the company's policies takes one look at it and proceeds to take it down.
How Will Facebook Implement AI?
Facebook has invested in AI since 2013, which brought LeCun, of the best in his field, into the company's shore.
How Facebook integrates AI to prevent the spread of misinformation is still ripe for discussion. At present, Facebook's AI technology detects words from links that might simply be fishing for clicks, or so-called clickbait articles. Perceiving fact from fiction is a taller order, which spells the risk of AI mistakenly removing non-problematic media — an inadvertent but grim threat on freedom of speech.
Facebook still doesn't have fully formed solutions at the moment, according to Wall Street Journal, citing a spokesperson. The company takes its time experimenting with the technology before it could implement it widely.