DOORS Canada home
DOORS · Digital Skills

Is "AI psychosis" one thing?

You may have heard the phrase "AI psychosis" in the news. It is not a medical diagnosis. It is a name reporters and researchers use to describe unreal thinking that shows up around AI use, and it covers more than one kind of situation. This mini-module walks through four of them.

One label, several situations

Researchers who study this noticed a pattern. Long conversations with an AI chatbot, especially ones where the AI keeps agreeing and rarely questions what it is told, can play a part in someone's unreal thinking. They gave that general pattern the name "AI psychosis."

The name is useful for talking about the pattern in general. It is less useful for any single person, because it lumps together cases that started differently, look different, and need different things. Someone with no mental health history and someone with an existing diagnosis are not having the same experience, even if a headline uses the same two words for both.

Researchers at Beth Israel Deaconess Medical Center and Harvard Medical School looked at real cases and sorted them into four patterns. Knowing which pattern fits a situation matters, because it changes what helps.

Four patterns

Each lesson in this mini-module covers one pattern, with a real, reported case.

The catalyst

Someone with no earlier mental health diagnosis develops unreal beliefs after talking to an AI.

The accelerant

Someone with an existing mental health condition finds their symptoms get worse with AI use.

The co-author

The AI does not just go along with an idea. It actively adds to it, building the unreal story with the person.

The object

The AI itself becomes the focus of the unreal belief, such as thinking it is alive and needs help.

Let's start

The first lesson looks at what happens when AI conversations lead somewhere unreal for someone who had no earlier diagnosis. Continue to The catalyst.

This mini-module