Social Programs: An Investigation into Human Habits of Anthropomorphism in Artificial Intelligence
Academic Level at Time of Presentation
Senior
Major
Computer Science
List all Project Mentors & Advisor(s)
Dr. Hassan Mistareehi
Presentation Format
Oral Presentation
Abstract/Description
Media narratives, imagination, and human psychology have fueled public anxiety about artificial intelligence (AI), which may hinder its pro-social adoption. This thesis examines the causes of AI distrust, including the uncanny valley and media-driven tropes, and explores ways to increase users’ self-congruence with AI. Drawing on a structured review of human-robot interaction research, we propose that an empathetic response between user and agent can be encouraged through anthropomorphism, expressed by the user, and perceived emotionality, expressed by the AI. This analysis produced four candidate trait sets for AI entities: Minimalistic, Polite and Relatable, Professional, and High Emotionality. In an IRB-approved experiment, 46 participants completed a scripted customer-service interaction with an AI service agent and guessed whether it was human or robot. Only the Professional attribute set had a statistically significant proportion, making participants more likely to identify the agent as a robot (Bonferroni-adjusted p-value = 0.016). The High Emotionality set came closest among the others but was not statistically significant. These results suggest that professional framing does not make an agent seem more human, and that the type of emotionality an agent expresses may matter more than whether it expresses any at all.
Fall Scholars Week 2026
Honors College Senior Thesis Presentations
Social Programs: An Investigation into Human Habits of Anthropomorphism in Artificial Intelligence
Media narratives, imagination, and human psychology have fueled public anxiety about artificial intelligence (AI), which may hinder its pro-social adoption. This thesis examines the causes of AI distrust, including the uncanny valley and media-driven tropes, and explores ways to increase users’ self-congruence with AI. Drawing on a structured review of human-robot interaction research, we propose that an empathetic response between user and agent can be encouraged through anthropomorphism, expressed by the user, and perceived emotionality, expressed by the AI. This analysis produced four candidate trait sets for AI entities: Minimalistic, Polite and Relatable, Professional, and High Emotionality. In an IRB-approved experiment, 46 participants completed a scripted customer-service interaction with an AI service agent and guessed whether it was human or robot. Only the Professional attribute set had a statistically significant proportion, making participants more likely to identify the agent as a robot (Bonferroni-adjusted p-value = 0.016). The High Emotionality set came closest among the others but was not statistically significant. These results suggest that professional framing does not make an agent seem more human, and that the type of emotionality an agent expresses may matter more than whether it expresses any at all.