Principal Investigator
Ying Xu – Harvard Graduate School of Education
Project Description
This study evaluates the impact of artificial intelligence (AI)-enhanced interactive science, technology, engineering, and mathematics (STEM) media on children’s learning, engagement, and identity development. Conducted with approximately 200 children four to eight years old in home settings, the randomized controlled trial (RCT) compares children who view interactive episodes with embedded conversational AI to those who watch standard broadcast versions of the same content. The AI features are designed to simulate caregiver-like interactions by prompting children to respond, reflect, and engage in problem-solving during the viewing experience. The study measures outcomes including STEM knowledge, transfer of learning, problem-solving skills, and STEM interest and identity, both immediately after viewing and one month later. The research also examines variation in effects across student characteristics and engagement patterns to understand how AI-enhanced media may support more equitable access to high-quality STEM learning experiences.
Research Questions
- What is the impact of AI-enhanced interactive STEM media on children’s learning outcomes compared to standard media?
- What is the impact of AI-enhanced media on children’s STEM interest and identity development?
- How do the effects of AI-enhanced media vary across children’s characteristics, including demographics and baseline STEM interest?









