The social network has been an excellent platform for mobilizing protests. However, once a demonstration reaches the streets, activists and authorities often clash. While organizing a protest on Twitter is arguably a noble thing, often causes get lost as demonstrations turn violent. A research team out of the University of Southern California’s Brain and Creativity Institute is using AI to help solve a problem. Specifically, artificial intelligence is predicting when organized protests will turn violent. The team used machine learning techniques to study 18 million tweets that were sent during the 2015 Baltimore protests. Demonstrators took to the streets to protest the coma and later death of Freddie Gray at the hands of police. The racially charged incident turned into violent pockets in the protest. “Our findings suggest that people are more likely to condone violent protest of an issue when they both see it at as a moral issue and believe others share this position, a pattern we refer to as moral convergence,” Joe Hoover, a USC PhD student who led the study, told Digital Trends. Researchers investigated the arrest rate during the demonstrations and tweets with a moral indicator. Arrests are a good way to gauge the general violence and misconduct, while tweets related to issues offering a right or wrong judgement can gauge user reaction. Interestingly, the team found the number of arrests during the protest matches the number of moral tweets before the demonstration.
Results
The team says people who pass moral judgement leading up to a protest are more likely to endorse violence or be violent themselves. Researchers describe an echo effect where moral tweeters believe other people share their views. “By tracking moralized tweets posted during the 2015 Baltimore protests, we were able to observe that not only did their volume increase on days with violent protests, but also that their volume predicted hourly arrest rates, which we used as a proxy for violence, during the protests,” Hoover said. “To further unpack these effects, we conducted a series of controlled behavioral experiments and we consistently observed the same effect of moral convergence.”
