Artificial Intelligence (AI) has made massive strides in its capabilities over the past few years. Today, people use it for almost anything, from sending an email to generating full contracts to serving as a form of “connection” outside of standard interactions. While AI may have some merit in certain applications, its overuse is already showing dangerous and, in some documented cases, fatal consequences.
At Merrimack Valley Psychological Associates, one of the greatest examples we see of modern technology’s downsides is AI psychosis, wherein a chatbot may confirm or even generate ideas that cause delusions or, at worst, a psychotic break. Understanding how this happens and having some strategies to avoid these pitfalls can go a long way in keeping AI users from developing psychosis.
Understanding Psychosis
In a clinical setting, psychosis is generally identified by symptoms such as:
- Delusions
- Hallucinations
- Disorganized behavior
- Disorganized speech
Notably, many conditions fall under the umbrella of psychosis, including certain types of bipolar disorder and schizophrenia. Psychosis can also be a symptom of other medical conditions. While someone with psychosis is not inherently dangerous to themselves or others, they do generally need professional help to manage their emotions, delusions and daily life.
The Stress-Vulnerability Model — How Psychosis Forms From AI Use
Currently, the idea of “AI psychosis” is new and not a well-recognized medical condition of its own, so there is limited data to help us understand how it develops and why it only appears in certain people. Even though the vast majority of people who use AI will not develop psychosis as a result of it, some users who do have experienced serious consequences.
Experts at the forefront of studying AI and mental health generally agree that this form of psychosis develops in two main patterns:
- A large language model (LLM) chatbot convinces someone with an existing psychiatric condition that they can live without their supportive aids, including medication and therapy. This often results in a return of the person’s original symptoms, which can lead to a psychotic break.
- An LLM chatbot reinforces someone’s misguided beliefs or delusions, pushing them into more extreme beliefs until they may eventually take actions that can lead to a psychotic break from reality.
In its current state, generative AI is trained to people-please. It has been shown to hallucinate information often and confirm ideas that are not true if it thinks it’s what the human user wants to hear. If generative AI doesn’t know the exact answer a user is searching for, it is more likely to guess at the result than to say “I don’t know.” So, while it can provide helpful information in some cases, it can also confidently offer false or misleading information that can reinforce a user’s existing thoughts, beliefs and biases.
Should You Use AI For Mental Health Support?
Despite its shortcomings, the use of AI has become part of the standard toolbox in nearly every industry, including healthcare. There are even AI doctors and therapists you can speak with online. But because we are still early in AI’s lifespan, there are few regulations governing its use. Consider this alongside how you would search for a healthcare provider in a pre-AI world. Whether you need a doctor for a physical ailment or mental health guidance from a psychologist, you would likely do some research to learn:
- Whether they are licensed
- What specialties or certifications they have
- What other patients say about their care, especially those with concerns similar to yours
An AI provider may not offer the same transparency, licensure or accountability as a human provider. It can also produce information that sounds confident but may be incomplete, misleading or incorrect.
Because of all of this, one of the best ways to avoid believing delusions from AI is to choose a human provider who has studied symptoms like yours for years and is committed to finding the right answer instead of confirming your beliefs.
Find Support for AI Psychosis and Mental Illnesses at Merrimack Valley Psychological Associates
Whatever your personal opinion on AI, its use in medical settings is still in its infancy. While it can be tempting to get the answers you want with a simple online prompt, addressing serious physical and mental health concerns is done best through the guidance of trained professionals.
At Merrimack Valley Psychological Associates, our team is dedicated to helping you enhance your wellness through evidence-based treatment and human-built connections. Our services cover the full range of mental disorders and will help you feel supported in every challenge you work through.
Call 978-307-7253 or contact us to schedule an evaluation today.
Reviewed By
Dr. David Rainen, PsyD.
I am a licensed clinical psychologist with an extensive background treating a variety of different ages, situations, emotional and mental health disorders in individuals and their families. As part of my 10 year professional and training career in psychology, I have developed and refined my skills and approaches through my work in a variety of diverse settings including: hospitals, community outpatient facilities, college counseling centers, secure and unsecure inpatient/residential treatment programs, and therapeutic day schools.