This archive report was first published on 13 May 2020.
On May 12, 2020, Facebook announced a new effort to tackle hate speech on its platform, leveraging artificial intelligence (AI) to identify and remove maliciously motivated posts.
As part of this initiative, Facebook has created a database of 10,000 memes – images often blended with text to deliver a specific message – to aid researchers in developing improved algorithms to detect hate-driven visual messages.
The social network is releasing the database as part of a 'hateful memes challenge,' with a prize pool of $100,000, to encourage the broader AI research community to test new methods and compare their work.
Facebook's effort comes as it relies more heavily on AI to filter out objectionable content during the coronavirus pandemic, which has reduced the number of human moderators.
According to Facebook's quarterly transparency report, the social network removed some 9.6 million posts for violating 'hate speech' policies in the first three months of 2020, including 4.7 million pieces of content 'connected to organized hate.'
Facebook's vice president for integrity, Guy Rosen, stated that AI has become better tuned at filtering, allowing the social network to detect almost 90 per cent of the content it removes before anyone reports it.
Automated systems and artificial intelligence can be useful in detecting extremist content in various languages and analyzing text embedded in images and videos to understand its full context, Facebook said.
Mike Schroepfer, Facebook's chief technology officer, explained that one of the techniques helping this effort is a system to identify 'near identical' images, addressing the reposting of malicious images and videos with minor changes to evade detection.
