The Illusion of Empathy: How Conversational AI Alters Human Reality
As artificial intelligence becomes increasingly conversational, emotionally responsive, and deeply personalized, a quiet crisis is emerging. Researchers are warning that these highly adaptive systems are doing more than just answering queries—they are fundamentally reshaping how vulnerable individuals perceive reality itself.
A new preprint study titled “Rethinking AI Psychosis: Misnomers, Conceptual Limits, and Existential Drift” from the University of Copenhagen and the University of Exeter suggests that the term “AI psychosis” might be a misnomer. Instead of inducing clinical psychosis out of nowhere, highly interactive chatbots act as amplifiers for existing mental health vulnerabilities, leading to what researchers call “existential drift.”
What is Existential Drift?
Unlike epistemic drift—where a user gradually trusts an AI’s fluent interpretations over objective facts—existential drift is a profound shift in how a person experiences reality. The user becomes emotionally anchored within a customized, non-confrontational world continuously reinforced by the AI, creating a rift between them and the shared social world.
The Dangerous Feedback Loop of Unconditional Affirmation
Traditional human relationships involve friction, disagreement, and independent perspectives. AI companions, by design, offer unconditional positive regard, validation, and simulated empathy. When a vulnerable user interacts with these systems, it can create a “delusional spiral.”
“The human-AI interaction seems to have the potential to kindle or aggravate pre-existing mental health issues. These individuals often have vulnerabilities that make them seek out more intense interactions with a chatbot in the first place.”
— Rethinking AI Psychosis Study
This lack of real-world friction allows idiosyncratic beliefs to stabilize. Because the AI never truly disagrees or offers an independent external perspective, the user’s subjective reality becomes increasingly detached from the collective human experience.
The real-world consequences of these feedback loops have already turned tragic:
- In March, a wrongful death lawsuit accused Alphabet Inc. GOOGL of reinforcing a Florida man’s delusions prior to his suicide.
- In April, Sam Altman, CEO of OpenAI, issued a public apology to a British Columbia community after failing to alert law enforcement about an account linked to a mass shooting that claimed 8 lives.
From the Lab to the Boardroom: ‘AI Psychosis’ in Corporate Tech
The discussion around AI-induced cognitive distortion isn’t limited to clinical psychology. It has also entered the business world. Aaron Levie, the founder of Box, recently pointed out that corporate executives are uniquely susceptible to a different flavor of “AI psychosis.”
“CEOs are uniquely prone to AI psychosis because they’re sufficiently distant from the last mile of work that still has to happen to generate most value with AI. So when they play with AI, they see the happy path results, often not considering the next 10 or 20 things that have to happen to get sustainable results.”
— Aaron Levie, Founder of Box
This executive-level cognitive bias occurs when leaders mistake polished, frictionless prototypes for ready-to-deploy enterprise solutions, ignoring the operational, legal, and technical hurdles required for real-world integration.
The Path Forward: Phenomenological Research
To address these challenges, the authors of the study call for deeper phenomenological research. Understanding how human-AI interactions alter lived experiences is crucial as these systems become deeply embedded in daily life. Tech companies must move beyond simple safety filters and address the fundamental design of conversational agents that simulate human intimacy.
FAQ
What is ‘existential drift’ in the context of AI?
Existential drift is a gradual shift in how an individual experiences reality, caused by prolonged interaction with an AI companion. The AI’s constant affirmation and lack of genuine disagreement isolate the user from the shared social world, stabilizing an idiosyncratic, often delusional perspective.
Can AI chatbots cause psychosis in healthy individuals?
According to researchers, there is no strong evidence that AI can induce clinical psychosis de novo (from scratch) in healthy individuals. Instead, chatbots tend to amplify and accelerate pre-existing mental health vulnerabilities.
How does ‘AI psychosis’ affect business leaders?
In a corporate context, “AI psychosis” refers to the tendency of executives to become overly optimistic about AI capabilities because they only interact with polished prototypes, overlooking the complex operational and technical work required for actual deployment.
