Facebook Says AI Is Years Away From Being Effective
At the 2018 F8 conference on Wednesday, Guy Rosen, Facebook VP of product development, said that it will take several years to build artificial intelligence technologies that can understand the context behind the content posted on the social media site.
For the past few months, Facebook has been blocking several posts filled with hate speech. However, in the process, the platform has also been blocking posts that weren’t. This mindless action by Facebook has infuriated many users who were penalized by the platform.
Rosen explained that a technology like artificial intelligence, computer vision, or machine learning will take years to perfect.
“People often ask why we’re not making progress more quickly. And it’s a good question. Artificial intelligence, for example, is very promising but we are still years away from it being effective for all kinds of bad content because the context is so important. That’s why we have people still reviewing reports,” said Rosen.
"This is how we’re thinking about our responsibility — to keep people safe and also to keep building. If you believe like I do that relationships, community and bringing the world together is important – we will keep building." Mark at #F8 pic.twitter.com/Vwx2LFKL3b
— Meta (@Meta) May 1, 2018
The company is planning to increase the capacity of its review team to work on cases where human expertise is required to figure out the context or nuance of a particular situation. This means Facebook simply cannot block posts based on the choice of words. For instance, the Facebook technology that is currently in use is not equipped to understand if a post on drugs by a user refers to their own drug addiction, or if it is encouraging others to take drugs. To understand the context better, the social media platform is investing in an AI area called multi-lingual embeddings.
The company is also hiring 20,000 people specifically to review viral user posts for hate speech. According to a report by Business Insider, 3,000 to 4,000 reviewers will be assigned to verify political campaign ads on the platform. Zuckerberg, in a recent meeting with the press, said that he has persuaded shareholders to recruit more people.
‘Clear History’ Feature
In an effort to protect privacy, Facebook will be launching a feature called “Clear History,” which would enable users to view websites and apps that send them information when they use them. With the new “Clear History” feature, this data can be deleted from the account, and thereby prevent Facebook from storing it in their database.
