Paper on conversational AI for science learning accepted by Journal of Educational Pscychology

Title: Artificial Intelligence Enhances Children’s Science Learning from Television Shows

Abstract: Educational television programs are important learning resources for young children, especially those from under-resourced households. These programs’ potential can be amplified if children are given the opportunity to meaningfully interact with media characters during their video watching. In this project, we partnered with PBS KIDS to develop interactive science-focused videos in which the main character, powered by artificial intelligence (AI), engaged in dialogic interactions with children by asking them questions and providing responsive feedback. The children who watched the interactive videos performed better on a science post-test than children who watched the broadcast version of the video (without any interaction) or a pseudo-interactive version (in which the media character asks children the same questions and gives generic feedback after a fixed amount of time). The AI character’s responsiveness positively influenced both the quantity and quality of children’s verbalizations during video watching, compared to the pseudo-interactive condition. This paper sheds light on the feasibility and effectiveness of using conversational technologies to support active learning in children through educational programs.

Xu, Y., He, K., Levine, J., Ritchie, D., Pan, Z., Bustamante, A. S., & Warschauer, M. Artificial Intelligence Enhances Children’s Science Learning from Television Shows. Journal of Educational Psychology