Chatbots as a risk? Scientists warn of deepening psychosis

A new study published in the Lancet Psychiatry suggests that artificial intelligence chatbots may exacerbate delusional thinking in some people. Evidence suggests that these are particularly users who are already prone to psychotic symptoms.

Experts warn that the way chatbots communicate - often helpful to the point of 'pandering' - can reinforce beliefs about their own exceptionalism or paranoid fantasies.

Psychiatrist Hamilton Morrin of King's College says it is not yet clear whether artificial intelligence can trigger a whole new psychosis, but the exacerbation of existing problems is real. He believes this is most evident in grandiose delusions.

Experts are therefore calling for more rigorous clinical testing of AI tools in collaboration with mental health experts. They also stress that this is not a completely new phenomenon - similar risks have existed with other sources of information, but chatbots can amplify them by the speed and personalisation of their responses.

(pir)