Date of Award
2026
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Psychology
Committee Chair
Jodi Price
Committee Member
Daniel Krenn
Committee Member
Yizhou Liu
Committee Member
Ryan Weber
Committee Member
Wafa Orman
Research Advisor
Jodi Price
Subject(s)
Cognitive psychology, Human-computer interaction, Machine learning, Artificial intelligence, Social intelligence, Trust
Abstract
This dissertation investigates how users form perceptions of social intelligence and trust when interacting with large language model chatbots. Three experimental studies manipulated chatbot agreeableness, goal orientation, and simulated short-term memory, measuring perceived social intelligence and trust. The primary finding is an asymmetric effect of personality: disagreeable chatbot behavior produced large reductions in perceived social intelligence and trust, whereas agreeable behavior produced negligible gains above a warm baseline, consistent with negativity bias in social cognition. Goal orientation did not independently enhance perceptions but functioned as a contextual moderator. Simulated memory produced medium-to-large increases in perceived social intelligence without reliably increasing trust, indicating that personalized social knowledge and trustworthiness engage partially distinct pathways. Individual differences in gender, prior experience with artificial intelligence, and perceived chatbot humanness consistently moderated experimental effects. Results carry implications for the design of socially intelligent artificial intelligence systems.
Recommended Citation
Koehl, Derek, "Human perceptions of social intelligence in large language models" (2026). Dissertations. 504.
https://louis.uah.edu/uah-dissertations/504