Much has been claimed about the benefits and harms of artificial intelligence (AI). Most of these claims are pure speculation, hype, or doom-laden pessimism. The scientific reality is that we have no real idea of how effective AI may, or may not, be in any domain—and there is certainly not enough basis to introduce it as a tool in anything so serious as healthcare. Keeping in mind those scientific caveats, it is worth highlighting a phenomenon that has received recent documentation and description—AI psychosis.

It has been noted that some individuals who engage with AI tools report symptoms of psychosis—especially delusions—resulting from their interactions with AI.1,2 It may not only be the AI interaction, but exposure to its products, such as deep fakes, that are implicated. In fact, any AI-generated doubt concerning the bedrock or background belief system that stops a cascade of delusional thinking from starting3 may be implicated.

Psychosis is a complex clinical phenomenon, involving multiple disruptions to an individual’s functioning, but a key aspect can be the presence of delusions. A standard definition of a delusion, which is a good working start, although not without some problems,3 is that a delusion is a belief that is false, often rare in the culture, and maintained despite strong evidence and consensus suggesting its implausibility. These types of delusions are often found in paranoid-type psychoses or schizophrenic disorders, but are not only restricted to those conditions.

Several reports have described how delusions appear to be associated with AI.1,2 One recent report1 describes a clear pathway for the........

© Psychology Today