For Immediate Release: April 15, 2023

Contact:
Tony Pals, tpals@aera.net
(202) 238-3235, (202) 288-9333 (cell)

Marla Koenigsknecht, mkoenigsknecht@aera.net
(202) 238-3233, (517) 803-1591 (cell)

AERA23 Study Snapshot: Interaction With a Television Character Powered by Artificial Intelligence Promotes Children's Science Learning

Study: “Interaction With a Television Character Powered by Artificial Intelligence Promotes Children's Science Learning”
Authors: Ying Xu (University of Michigan), Julian Levine, Valery Vigil, Daniel Ritchie (University of California – Irvine), Shan Zhang (Harvard University), Trisha Thomas, Carlos Barrera, Michelle Meza, Andres Sebastian Bustamante, Mark Warschauer (University of California – Irvine)
Embargoed until: 12:01 a.m. CT Saturday, April 15

This paper will be presented at the place-based component of the AERA 2023 Annual Meeting. 
Session: Integrating Computer Science and Artificial Intelligence in Teaching and Learning
Date/Time: Saturday, April 15, 8:00 a.m. – 9:30 a.m. CT

Main Findings:

• This project showcases a collaboration between PBS Kids and university researchers to integrate artificial intelligence into children’s television shows so that the shows’ characters can ask children questions, comprehend children’s responses, and provide specific feedback.
   
• Offering children opportunities to converse with an AI-powered interactive media character in a science video leads to better learning outcomes than for children watching videos with no interactions or those with “pseudo” (not fully two-way) interactions that merely ask children questions and pause to allow them to respond (as is done on Dora the Explorer).

Details:

• Children between the ages of 3 and 6 spent about two hours daily, on average, watching television or videos. Children’s engagement with and comprehension of educational content can be further promoted by adding in-the-moment interaction to the programming.
   
• The authors examined the effect of an AI-generated interactive character in the videos by using a randomized controlled experiment involving 240 children. The participants ranged from preschool to first grade, predominantly of Hispanic and Latino ethnicity.
   
• The authors collaborated with PBS KIDS to develop conversational videos adapted from a popular children’s science animation show, Elinor Wonders Why. They used two conversational videos corresponding to two episodes.
   
• The children were assigned to one of three conditions: conversational (watching the conversational video and having contingent interactions with the show’s main character); pseudo-interactive (watching the video with pseudo-interactions where the character paused occasionally during the video to ask the same questions, but the character gave generic feedback after a fixed amount of time); and non-interactive (watching the broadcast version of the video without any interaction).
   
• The study consisted of four sessions. In the first two sessions, the authors pretested children’s English and Spanish skills and science knowledge. In the third and fourth sessions, the children watched the first and second episodes in their assigned format and completed a posttest on each episode. The children were surveyed about their perceptions at the end of the third session.
   
• Children who watched the conversational version of the video performed the best in the post-viewing assessment (correctly answered 63.4 percent of the questions) followed by children who watched the pseudo-interactive version (correctly answered 61.4 percent) and the non-interactive version (correctly answered 56.5 percent).
   
• The results also tentatively suggested that neither age nor previous science knowledge were significant moderators for the assigned video condition.
   
• “The results of this study suggest that interactive videos are most effective in ‘breaking the fourth wall’ to engage children in the learning process,” said study coauthor Ying Xu, an assistant professor of learning sciences and technology at the University of Michigan.
   
• Children who watched the interactive and pseudo-interactive videos reported a more positive perception of the program and the character than those who watched the non-interactive videos. There was no significant difference in perception between the interactive and pseudo-interactive versions.
   
• “As children spend more time watching videos and shift to Internet-connected devices, it is imperative to investigate how new forms of video watching may better support learning,” said Xu. “Research and design communities should take advantage of evolving conversational technologies to maximize the educational benefit of children’s screen time.”

To request a copy of the working paper, or to talk to the study author, please contact AERA Communications: Tony Pals, Director of Communications, tpals@aera.net, cell: (202) 288-9333; Marla Koenigsknecht, Communications Associate, mkoenigsknecht@aera.net, cell: (517) 803-1591.

###

About AERA
The American Educational Research Association (AERA) is the largest national interdisciplinary research association devoted to the scientific study of education and learning. Founded in 1916, AERA advances knowledge about education, encourages scholarly inquiry related to education, and promotes the use of research to improve education and serve the public good. Find AERA on Facebook, Twitter, LinkedIn, and Instagram.