Design GenAI for
Social Media Literacy Training

From online harassment to misinformation, anti-social behaviors are alarmingly prevalent on social media and can cause serious harm. One proven antidote is prosocial users, for example, citizens who vigilantly counter misinformation, and bystanders who intervene after cyberbullying by condemning the behavior and supporting the victim.

We design novel LLM-powered social simulations to cultivate prosocial social‑media behaviors among late adolescents and young adults. The HAI research question underlying this work is: How can we best harness LLMs’ language abilities to foster user reflexivity, while minimizing the risk that LLMs go rogue amidst online misinformation, toxicity, and aggression?

This project is ongoing. Below are some initial publications:

• A Piece of Theatre: Investigating How Teachers Design LLM Chatbots to Assist Adolescent Cyberbullying Education (Hedderich et al., CHI’24)

This project is supported by the NSF. Jessie Jia is its student lead. Alumni Michael Hedderich and Nader Akoury were previous student leads.