Electronic Theses and Dissertations

Jingjing Wu

Date

2025

Document Type

Dissertation

Degree Name

Doctor of Philosophy

Department

Business Administration

Committee Chair

Huigang Liang

Committee Member

Srikar Velichety

Committee Member

Yafang Li

Committee Member

Yajiong Xue

Abstract

Amid growing concerns about loneliness as a public health crisis, artificial intelligence (AI) technologies have emerged as promising tools for providing socio-emotional support. Yet, why such technologies influence user loneliness remains underexplored. This dissertation examines whether voice-based AI can help users feel less lonely and explore the mechanism. It further explores what demographic will be impacted more to which AI voice design. Anchored in Self-Determination Theory and Computers Are Social Actors (CASA) paradigm, the dissertation focuses on a key psychological construct: perceived empathy. Using self-designed voice-based AI, two large-scale randomized experiments (N₁ = 195; N₂ = 746) are conducted interactions on CloudResearch Connect subject pool platform. Participants conversed with the AI companions whose design varied systematically in anthropomorphism (high vs. low), voice gender (male vs. female), and voice age (younger vs. older). Results consistently showed that anthropomorphic AI significantly reduced loneliness through heightened perceptions of AI empathy. Mediation analyses confirmed that perceived empathy mediates the relationship between anthropomorphism and loneliness. Furthermore, moderated mediation analyses revealed that this effect was not uniform across population: it was significantly stronger among female users, and particularly so when interacting with male-voiced or younger AI agents. These findings suggest that both user demographics and AI voice features shape the psychological benefits of AI interactions. Theoretically, this research extends the Computers Are Social Actors (CASA) paradigm by demonstrating that voice-based AI can elicit authentic emotional and social responses, moving beyond surface-level human mimicry. It further introduces Self-Determination Theory (SDT) into the human–AI interaction literature, showing that AI can fulfill users’ psychological need for relatedness—reducing loneliness in ways comparable to human connection. Practically, the findings provide clear design implications for developers of voice-based AI: customizing AI voice characteristics (e.g., gender, age) and enhancing empathic responsiveness meaningfully improves user experience. These design elements not only increase engagement but also boost AI’s effectiveness as a tool for emotional support. The results are particularly valuable for applications in healthcare, eldercare, and mental health services, where voice-based AI can serve as a scalable, low-cost, and accessible intervention to address chronic loneliness. Overall, this work offers an initial foundation for designing and deploying emotionally intelligent AI systems that promote psychological well-being. Limitations and future directions are discussed at the end.

Comments

Data is provided by the student.

Library Comment

Dissertation or thesis originally submitted to ProQuest.

Notes

Embargoed until 08-04-2027

Download

Available for download on Wednesday, August 04, 2027

Share

COinS