Speaking Like Someone You Know: Exploring Transference in Human-AI Voice Interactions

This proposed research explores how synthetic voices that mimic familiar individuals influence user perceptions and behaviors in human-AI interactions. Grounded in transference theory, we hypothesize that customized voices resembling significant others can evoke a sense of closeness and affect task performance by activating the mental representations of those individuals. The proposed research investigates how factors such as emotional bonds, the need for interaction, and attitudes toward AI moderate these effects. Our study aims to test multiple hypotheses about closeness and performance outcomes through a lab experiment involving anagram-solving tasks with customized or generic AI voices. This work contributes to the literature on human-computer interaction, transference, and anthropomorphism by showing how personalized audio can shape cognition and emotional engagement.