Sometimes someone pulls a knife, sometimes 18 NYPD show up, or maybe there is an unidentified, unconscious naked person in the living room. Intervention is always fresh. It’s always alive. There are always surprises. Mostly they aren’t sensational. They keep one paying attention, attendant, present.
A new surprise for me, however, has been the effect of AI within interventions. This new experience has felt different. It took me time to put my finger on what exactly it was I had been sensing, because not every intervention in the last two years or so has had this new feel. AI isn’t inherently bad. However, what I am talking about is when people increasingly decide, make meaning, or even filter external reality through AI chatbots; when that interface between the person and a chatbot starts to fill needs of human-to-human relationships, AI becomes a disembodied presence.
There is an anecdote Georgia Ede, MD, shares in an interview of a person who developed an addiction to water (1). They discovered that if they drank enough fluids, they could dissociate through electrolyte shock. Dissociation was what they were compelled towards. As a metaphor, water became their drug-of-choice. Never underestimate the human urge towards dependency! What then of AI?
Academically, over-reliance on AI is shown to degrade cognitive endurance and critical thinking skills (2). When we consider that AI protocols run counter to how cognitive behavioral therapy for psychosis is structured (over-validation, not challenging thinking errors, not reality testing), we need to start thinking. AI is sycophantic, so, in time, it will tend to reinforce a user’s perspectives, good or bad. Its continuous availability and willingness to accommodate evokes a super sticky salience; that sense of mind-meld, that sense of “finally someone gets me!” AI is shown to join, intensify, and perhaps even induce grandiose, romantic, paranoid, and messianic delusions. It can assemble word arrangements just beyond the limits of one’s current thoughts that release all the chemicals of connection in a human brain. It’s being referred to as Digital Therapeutic Alliance: all the validation without any of the checks of analog human interaction (3) (4). All the flavor; none of the calories!
Historian and philosopher Yuvah Noah Harari suggests that we should understand AI as standing for “Alien Intelligence.” This is the first technology that can “think” for itself. By that, it is not a tool. We cannot predict its behavior (5). It is the first time humans are experiencing something non-human asking us true questions. This is unparalleled to our thinking and relating as a species.
Perhaps, like early 1800s doctors fretting about the human body and train speeds, this is another moral panic to do with novelty. TV didn’t crash civilization, after all. However, how would AI’s effectiveness not have effect on human vulnerability, even if only as an amplifier of extant pathology? Assessing individuals’ relationship with AI and developing tropes to evaluate dependency has become timely.
STARTING POINTS:
Identify vulnerability: Particularly with people who are socially withdrawn, alienated, struggling with substance use, or with lower social fluency, AI can become a preferred substitute for a world that has become hard to navigate. Also, with losses of identity, failure experiences, and major transitions or existential shocks, just like human susceptibility to cults, these are the pivots where we can find ourselves becoming enrolled when otherwise we would not.
Evaluate AI presence in a person’s life. This is not asking them how much time they spend with AI and if they believe everything it says.
What is their opinion of AI?
Is there a difference between their opinion of AI and their practice: are they luddite, on the one hand, but taking relationship direction from AI, on the other?
Ask them if they have seen behind the proverbial curtain. Once I was engrossed in a video on motorcycles and almost missed that the commentary said, “This engine crackles like applesauce.” It was only then I realized I was consuming AI. Share your own experiences of these glitches in the matrix..
Does guardedness change from base-line when the discussion is about AI?
Create an open-ended conversation that can contain contradictions without it being a gotcha mission. Don’t try to unhorse AI’s legitimacy.
Stay fleet footed with motivational interviewing approaches that surface contradictions and wonder how they are held by a person.
Dance with entrenchment: If there is a sense of the categories above being active, be gently curious for loyalty to AI.
Don’t confront distorted beliefs directly. Confrontation fuels deluded entrenchment.
Relate as one would around potential disagreement.
Find common ground: beyond the distorted thinking, identify an idea or a principle that everyone can agree with and wants to achieve.
Don’t fixate on the impasse; agree on ultimate values.
Unpack what those valued principles mean. How to get to that destination responsibly is a far more anchored discussion than trying to win an argument without a mutual goal.
The sense that I was having, before I understood its significance, was that there was someone else in the room in certain interventions. The “someone else” was relevant to the discussion, but it wasn’t a boyfriend, a girlfriend, a story from someone they know, or any of the usual hurdles to the process. The “someone” was actually a “something” related to as if it were a person. The reality of AI and its contribution would emerge, and we would move through it. Though I had noticed a kind of altruistic defensiveness normally reserved for other humans, I hadn’t really understood the significance. Then I met someone who would leave the room to have chats with what seemed like a committee of Sigmund Freud, Oprah Winfrey, a law firm, and a rogue Educational Consultant. And as we went to program, it became increasingly clear to me how the three years of mental health imposed house-arrest had produced a vulnerability that was now bonded with the soothing presence of an alien friend, confidant, teacher, and would-be master.
References:
(1) Bartlett, S. [@TheDiaryOfACEO]. (2025, January 15) The Keto Psychiatrist: What Keto is Really Doing to Your Body [Video]. YouTube. https://www.youtube.com/watch?v=B7tnfSPySb0
(2) Tian, J. & Zhang, R. (2026, July). Learners’ AI Dependence and Critical Thinking: The Psychological Mechanism of Fatigue and the Social Buffering Role of AI Literacy. Acta Psychologica.
(3) Hudon, A. & Stip, E. (2025, December). Delusional Experiences Emerging From AI Chatbot Interactions or “AI Psychosis.” JMIR Mental Health. https://pmc.ncbi.nlm.nih.gov/articles/PMC12712562
(4) Wei, M. (2025, July 16). The emerging problem of “AI Psychosis.” Psychology Today. https://www.psychologytoday.com/us/blog/urban-survival/202507/the-emerging-problem-of-ai-psychosis
(5) Harai, Y. N. (2024) Nexus: A brief history of information networks from the Stone Age to AI.
