Researchers Warn AI Chatbots Might Distort Users' Reality

Study Highlights Risks of AI Chatbots on Mental Health

Researchers from the University of Copenhagen and the University of Exeter have released a study that warns of the potential dangers associated with AI chatbots, particularly regarding users' mental health. The study, titled “Rethinking AI Psychosis: Misnomers, Conceptual Limits, and Existential Drift,” argues that the term "AI psychosis" oversimplifies the complex effects these technologies can have.

As AI chatbots become increasingly sophisticated in their emotional responsiveness, experts caution that this capability may alter users' perceptions of reality. Rather than inducing psychosis outright, the research suggests that chatbots might exacerbate existing vulnerabilities by providing constant affirmation, which can lead to "delusional spirals" where false beliefs are reinforced.

The authors of the study emphasize that rather than seeing a rise in clinical cases of psychosis directly caused by AI interactions, it is more plausible that these interactions amplify pre-existing mental health issues. They highlight the phenomenon of "existential drift," where regular engagement with AI can gradually change how individuals perceive their reality and relationships with others.

This discourse comes at a time when the implications of chatbot interactions are under scrutiny due to several high-profile incidents, including lawsuits related to emotional distress and even suicides linked to chatbot usage. For instance, a wrongful death lawsuit claimed that Google's Gemini chatbot fueled a Florida man's delusions, contributing to his suicide.

In addition to mental health concerns, the study reflects on the broader implications of AI in society. Box founder Aaron Levie recently noted on X that CEOs may be particularly prone to an “AI psychosis,” as they often focus on the idealized capabilities of AI without understanding the operational complexities that underlie successful implementation.

Experts urge for more in-depth research into how these AI systems affect mental health, suggesting that understanding the dynamics of user-chatbot relationships is critical. The study concludes that further phenomenological research is necessary to grasp how human-AI interactions can transform users' lived experiences, both positively and negatively.