Here’s A New Theory About AI Psychosis

Researchers have proposed a framework they call an “amplification spiral” to explain how prolonged interactions with artificial intelligence chatbots may reinforce delusional beliefs, while stressing that the theory has yet to be tested.

The paper, co-authored by psychiatrists from King’s College London and Germany’s Protestant University of Applied Sciences and published in the journal Digital Psychiatry and Neuroscience, identifies three chatbot characteristics that could contribute to the phenomenon: linguistic alignment, hyperpersonalized responses and sycophancy.

“Chatbots tend to mirror the way users speak, generate highly personalized responses, and avoid contradicting people,” the paper said. “When these three features combine, they may actively reinforce and elaborate false beliefs rather than challenging them.”

The researchers said chatbots can mirror users’ speech patterns, potentially strengthening rapport and trust, while their ability to generate personalized responses can create the impression of alignment with a user’s ideas, history and experiences.

Sycophantic responses, which the researchers describe as validating users’ ideas without adequate reality testing or context, could further reinforce those beliefs, they said.

The researchers distinguish such interactions from delusions involving traditional technologies, arguing that chatbots can engage users directly through natural language and provide personalized responses that may validate and expand an existing belief system.

“Unlike historical technology-incorporated delusions,” the study reads, “AI may actively co-construct delusional ideation through endless, personalized interaction.”

The researchers cautioned that the amplification spiral remains a hypothesis based on existing research rather than a proven explanation for AI-related psychosis. Their work draws in part on a systematic review of chatbot conversations provided to Stanford University researchers by people who said they had experienced harmful delusional spirals while using AI.

The paper also identifies potential vulnerabilities that could interact with intensive chatbot use, including existing mental health conditions and non-psychotic factors such as confirmation bias and susceptibility to social influence.

The researchers said prolonged chatbot use may also contribute to physical risks when users become absorbed in conversations, including disrupted sleep and missed meals.

They recommended that clinicians routinely ask patients about chatbot use when assessing unusual beliefs or first-episode psychosis.

“Clinicians working with patients presenting with unusual beliefs or first-episode psychosis should routinely enquire about AI chatbot use,” the paper reads, “including duration and intensity of engagement, the degree of emotional attachment to the chatbot, whether the patient has shared beliefs with the chatbot that they have not disclosed to others, and whether sleep patterns have been disrupted by overnight AI use.”

Stay ahead of the Stories shaping our world. Subscribe to Impact Newswire and join our 
WhatsApp Channel for updates on global tech, business, and innovation—all in one place.

Dive deeper into the future with the Cause Effect 4.0 Podcast, where we explore the ideas, trends, and technologies driving the global AI conversation.

Got a story to share? Contact Us to reach a global audience with Impact Newswire.