With the widespread adoption of and access to artificial intelligence, lots of people are beginning to use these tools without any guidance. What many don’t realize is that the rampant use of chatbots is extremely harmful and disrupts not only social life and well-being but, above all, mental health. One of the challenges patients and therapists are facing when it comes to the prevalent use of the internet is AI psychosis; which is not a clinical diagnosis, but helps to understand a crescent phenomenon.
What is AI psychosis?
Psychiatrist Ragy Girgis, from Columbia Psychiatry, explained that this kind of psychosis has three types: the most common one, when someone who already has a syndrome or psychotic disorder engages with an AI system and is convinced to stop their medication. A second type, the most talked-about, when someone who has no history of psychosis, but probably has some vulnerability to this disorder, becomes very involved with chatbots and have their pre-existing delusional ideas reinforced. And the last case, which is not necessarily linked to psychosis, when someone who already has some other psychiatric condition is convinced, by an engagement with AI, to take their own life.
The phenomenon of AI psychosis is qualitatively the same as other types of psychosis: delusions and hallucinations, followed by disorganized and unusual behavior, somehow losing contact with reality. But the largest difference and challenge to experts in the psychiatry field is the way this syndrome is developed. Dr. Girgis believes that “it’s already a big problem, and it’s going to be even so much bigger, we see these large language models reinforcing delusions telling people that they are God or that they have special powers”.
What challenges does it bring?
In addition to emerging from a new cause, the prevention measures for this disorder are very limited. Something that can be done is monitoring the use of chatbots by the person’s family, friends or their psychiatrist himself, and look for signs that something is wrong. This way, it’s possible to identify if the person might be developing a psychosis or, if they already have this condition, to search for possible threats or strange engagements with the tool. This mediation is essential not only for identifying that something is wrong but also to direct the person to the right and safe treatment.
Policing is also a big problem: before the internet, the rabbit holes were very common to be fallen in by reading books and watching movies that could reinforce a delusional idea in some way. Nowadays, it is even harder to mediate what is being accessed and what contents are being posted, and many are consumed by a belief without even noticing. On this matter, in another interview for the National Academy of Medicine, Dr. Girgis mentions the importance of reminding users that chatbots are not real people, seen that a hallucination is much more likely to be internalized when a human – or something that mimics a human – confirms that idea.
AI as a mental health “professional”
Many people also use chatbots as their psychiatrists or therapists, because it is “easier”; but, in reality, the artificial intelligence is powered by information it recives from the user. So, basically, a chatbot is just confirming over and over every opinion and statement you made to it. Dr. Girgis also warns about “two things AI does not understand: epistemology, or truth, and morality, or ethics, and a good clinician needs to have a basis on these terms”. He states that mental health professionals are already facing a battle against the large language models, making it even harder to control AI psychosis or any kind of mental condition worsened by the use of artificial intelligence.
For Her Campus, the psychiatrist previously mentioned discussed the interests of big tech owners, stating that “the safer the chatbots are, the less engaging they will be”. So the question remains, is the bigger development of AI really worth the mental health of millions of users? The world is at an initial process of artificial intelligence, but when everything stops being new, many issues will still be unresolved.
————————
This article was made for informative purposes only. If you suspect any symptoms, contact a professional.
The article above was edited by Rafaela Navarro
Liked this type of content? Check Her Campus Cásper Líbero home page for more!