Chenxi Wu

Topic

Can AI-Mediated Intergroup Contact Improve Intergroup Attitudes? A Contact Theory-Based Study of Social Chatbots

Department of Psychology

Date & location

• Monday, August 10, 2026

• 1:00 P.M.

• Cornett Building

• Room A228

Reviewers

Supervisory Committee

• Dr. Nigel Mantou Lou, Department of Psychology, University of Victoria (Supervisor)

• Dr. Kelci Harris, Department of Psychology, UVic (Member) 

External Examiner

• Dr. Alexandra Kitson, Department of Computer Science, University of Victoria 

Chair of Oral Examination

• Dr. Donald Juzwishin, School of Health Information Science, UVic

   

Abstract

As generative AI systems become increasingly capable of sustaining socially meaningful conversations, they create new opportunities for digitally mediated intergroup contact. The present study examined whether a brief, structured interaction with a culturally identified AI chatbot could improve majority-group students’ attitudes toward Chinese international students. Grounded in intergroup contact theory, the study randomly assigned European Canadian undergraduate students to one of two 15-minute chatbot conditions. In the experimental condition, participants interacted with a chatbot represented as a Chinese international student, using culturally meaningful self-disclosure prompts. In the control condition, participants interacted with DialogueBot, a chatbot without a specific cultural or ethnic identity, using parallel general friendship-building prompts. The sample included 76 eligible participants who passed the embedded attention check and correctly identified the chatbot’s intended background. 

The primary outcome was warmth toward Chinese international students, assessed before and after the chatbot interaction. Contrary to the primary hypothesis, the experimental condition did not produce significantly greater adjusted post-interaction warmth than the control condition, F(1, 73) = 0.08, p = .779. The pre-registered intercultural readiness outcome also did not differ significantly by condition, t(73.99) = -0.27, p = .787, d = 0.06. In secondary analyses, participants in the experimental condition showed a significant within-condition increase in warmth from pre- to post-interaction, t(42) = 2.09, p = .043, dz = .32, although the magnitude of warmth change did not differ significantly between conditions. Cultural interest showed a nonsignificant condition effect in the ANCOVA model, F(1, 73) = 3.23, p = .076, but change score analyses indicated a larger increase in Chinese cultural interest in the experimental condition than in the control condition, t(63.48) = -2.09, p = .041, d = 0.44. Participants also evaluated the culturally identified chatbot more positively than DialogueBot, rating it as higher iv in perceived interaction quality, anthropomorphism, and likability. 

Overall, the findings suggest that brief AI-mediated contact may not be sufficient to produce robust between-condition change in broad intergroup warmth. However, culturally grounded chatbots may still foster cultural interest and positive engagement, especially when participants perceive the intended cultural framing. AI-mediated intergroup contact may therefore be best understood as a low-stakes entry point into intercultural engagement rather than a replacement for direct human contact. 

Keywords: AI-mediated contact, chatbot, intergroup attitudes, intergroup contact theory, cultural interest, anthropomorphism, Chinese international students