The new system called WES (Web-Enabled Simulation) uses bots modelled on human-behaviour on Facebook.
California-based social media giant Facebook has developed a bot social network to detect potential vulnerabilities and pre-emptively address such threats before they affect real human lives. The bots replicate actual user behaviour and automatically interact with millions of other bots.
The company uses a combination of online and offline stimulation to train bots with skills ranging from simple rules like supervised machine learning to more sophisticated reinforcement learning. These automated software applications will not be able to interact with real Facebook users or affect the experience of actual humans on the platform.
The bots are trained to build social connections using the same infrastructure as real users, according to Facebook. They can conduct searches and visit pages, just like human beings.
[3 minute read]