Abstract

As AI chatbots become increasingly integrated into daily life, their influence on users’ psychological well-being demands urgent attention. While these systems offer benefits for learning, productivity, and companionship, they also present risks ranging from emotional entanglement and unhealthy dependence to the exacerbation of psychological vulnerabilities. In this talk, we first present findings from a research study that aims to understand how some people develop and recover from delusional beliefs about the nature of AI. We then present a set of aspirational directions for the design of general-purpose AI chatbots, with the goal of reducing psychological harm.

Speaker Bio

Forough Poursabzi

Dr. Forough Poursabzi is a Principal Applied Scientist at Microsoft. Her work lies at the intersection of society, people, and AI. Through an interdisciplinary lens and building on knowledge from fields of computer science, psychology, social sciences, and human-computer interaction, she works on identification and mitigations of risks in AI. More recently, her work has focused on understanding and identifying psychological risks of AI and translation of this research into evidence-based policy and design guidance. Forough's work informs responsible development of AI companions and chatbots, with an emphasis on safeguarding user wellbeing and promoting ethical human-AI relationships. Forough holds a PhD in Computer Science from University of Colorado Boulder and a BE in Computer Engineering from University of Tehran.

Amy Heger

Dr. Amy Heger is a social psychologist turned UX researcher/applied scientist supporting Microsoft’s Aether team (AI Ethics and Effects in Engineering and Research) through JeffreyM Consulting. She studies the psychological influences of AI, including overreliance, emotional connection, trust, transparency, and productivity. Her research shapes product team guidance, implementation frameworks, and organizational policy that help AI practitioners develop and deploy AI systems more responsibly. Before joining Aether, she was a Research Associate at Oak Ridge Associated Universities, where she designed safety culture assessments for nuclear facilities.

Mihaela (Mickey) Vorvoreanu

Dr. Mihaela (Mickey) Vorvoreanu is Principal Applied Scientist at Microsoft, where she co-leads the working group on Psychological Impacts of AI within Aether—the company’s advisory body on AI Ethics and Effects in Engineering and Research. Her work focuses on advancing the practice of responsible AI through evidence-based guidance, sociotechnical research, and human-centered design. Before joining Microsoft, Mihaela had a fulfilling academic career, including at Purdue University, where she founded and led both graduate and undergraduate programs in UX Design and Research.