Collaborative Research: HCC: Medium: Collaborative Upstanding: Leveraging Conversational AI to Cultivate Constructive Upstanders Among Teens

This project uses artificial intelligence (AI) to guide adolescents to stand up against bullying in social media. The approach encourages teens to do this in a safe and effective manner. The rate of cyberbullying in social media among teens is worrisome, and the potential consequences of cyberbullying can be detrimental. The prosocial behaviors of bystanders can play a critical role in combating cyberbullying. Such behaviors may include confronting the bully or comforting the victim. Reversely, bystanders’ inaction and silence allow bad actors to feel less inhibited and may result in a downward spiral of increasing aggression on social media. Using the latest AI technologies, the project aims to achieve human-AI collaborative upstanding. These AI technologies may include Chat GPT-3, AI chaining, and other approaches. Using these AI technologies, the project will provide personalized and empathetic guidance and result in the development of prosocial bystander behaviors by teens using social media.The investigators will first conduct empirical studies to identify successful collaborative upstanding strategies. These strategies will enable teens to encourage each other to stand up against bullies online. Investigators will also identify ways teens may fail in this process. Next, investigators will create a set of AI conversational agents and implement the identified successful conversational strategies. The system developed will go through an iterative prototype design process. Then investigators will perform an evaluation and improvement process to the prototype. Finally, the investigators will deploy and test the best performing conversational AI system. The project will test with youth the conversational AI using Social Media TestDrive. This is an online learning platform used by over 800,000 teens and educators in the United States.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.