“It can be really challenging for the average person to know where to start with taking meaningful action on social media, and we want to give people the tools to strengthen their voices and fight back against hateful posts online,” said Zachary Bamberger, a Cornell graduate now working toward a master’s in computer science at the Technion-Israel Institute of Technology.
Over the last month, he has enlisted the help of students, professors, Technion alums, and other volunteers to reinvent methods for combatting hate speech online.
This is a growing issue worldwide.
Ever since the Hamas killing and kidnapping rampage on October 7 that sparked a war with Israel, the number of anti-Semitic incidents, campus protests and online vitriol against Israel and Jews has skyrocketed.
The ADL reports that anti-Semitic incidents in the United States alone rose nearly 400 percent in the two weeks following the October 7 attack compared with the same period last year.
Bamberger’s alma mater in Ithaca, New York, had to cancel classes on November 3 after a student posted online antisemitic threats against Jewish people on campus.
Using machine learning and natural language processing, Bamberger built Rhetoric AI, which instantly generates high-quality translations of Arabic and Hebrew posts.
The large language model (LLM) predicts the virality of posts, reports posts that violate terms-of-service agreements, and produces counterarguments in Arabic, Hebrew or English for volunteers to utilize in their own responses.
“When deployed on a larger scale, our platform will shift the balance of content and take away the numbers advantage of those who propagate hate, violence and lies,” Bamberger said.
Rhetoric AI is already collaborating with some of the best-known experts in the field, including AI21 Labs, Google and Microsoft.