Facebook Uses AI To Fight COVID-19 Misinformation, Hate Speech; Here's How

Facebook is now using artificial intelligence (AI) to combat COVID-19 misinformation and take down hateful memes. The leading social media network unveiled the initiative on Tuesday in its aim to enforce strict community standards.

On Monday, Facebook issued the Community Standards Enforcement Report, which details how Facebook uses a combination of AI and human fact-checkers and moderators to moderate its platform amid the COVID-19 pandemic. Facebook also revealed putting warning labels on about 50 million COVID-19-related posts during the month of April to fight misinformation about the disease.

The social media network reported that their labels are working, as 95 percent of the time, people decide not to view a piece of content when they notice the misinformation warning. However, the social media network acknowledged that producing such labels across its enormous platform could be very challenging, even with the use of AI.

“Once independent fact-checkers have determined that an image contains misleading or false claims about coronavirus, SimSearchNet, as part of our end-to-end image indexing and matching system, is able to recognize near-duplicate matches so we can apply warning labels,” said Facebook. “This system runs on every image uploaded to Instagram and Facebook and checks against task-specific human-curated databases. This accounts for billions of images being checked per day, incuding against databases set up to detect COVID-19 misinformation.”

Facebook is now also using AI to curb hate speech on the networking site, which globally occurs in photos and videos. This is an even tougher challenge for AI, as software must also be trained to find duplicates or marginally modified versions of the content as it spreads across Facebook.

As of now, the network has already created a database of 10,000 memes as part of its ramped-up effort against hate speech. “These efforts will spur the broader AI research community to test new methods, compare their network, and benchmark their results in order to accelerate work on detecting multimodal hate speech,” said Facebook in its report.

Recently, Facebook also launched the “hateful memes challenge” to motivate researchers to develop improved algorithms that can detect hate-driven visual images. The first to create the algorithm will win a prize of $100,000.