Facebook researchers are developing AI-powered bots to tackle online harassment on platform .
According to the social media company's announcement, they hope the new technology will help in ongoing efforts to ensure the platform's artificial intelligence has the ability to eliminate harassment.
Now they can simulate the actions of users behaving badly, using AI-powered bots and letting them run free on a parallel version of Facebook. The researchers will then study the bots' behavior in the simulation and experiment with new ways to stop users with similar activity.
As researcher Mark Harman said , the simulator uses machine learning to train bots to simulate real-world human behavior on Facebook. It can also automate their interactions at scale, from thousands to millions, and is based on Facebook's code base
In the testing environment, known as WW, the bots perform various actions while attempting to buy and sell illegal items, such as weapons and drugs. The bots can use Facebook as a normal person would, performing searches and visiting pages.
Engineers can then test whether bots can bypass security and violate the Community Standards, according to the statement. The plan is for engineers to find a pattern in the results of these tests and use that data to test ways to make it harder for users to violate those standards.
Facebook has long said it is developing methods to prevent online harassment, criminal activity, misinformation and other types of illegality on its platform.
At Facebook’s 2018 F8 , Chief Technology Officer Mike Schroepfer said the company is investing heavily in AI research and looking at ways to make it work at scale with little or no human oversight, such as with bots. To Facebook’s credit, this simulation seems to be evidence of that.
