You ask the question. The answer calms you down. Twenty minutes later you are asking again, worded a little differently, just to be sure.
A lot of people now bring their anxious questions to AI the way they used to search them at 2am. The chat window is always open. It never gets tired of you. It answers instantly, patiently, and as many times as you want. That can feel like support. If you are asking to feel reassured, though, you may be feeding the worry instead of settling it.
Reassurance works. That is the problem. You get the answer, your body relaxes, and the anxiety drops. For a minute.
Then the doubt creeps back. Maybe the answer missed something. Maybe you did not phrase the question right. So you ask again. That short burst of relief is exactly what makes it so easy to keep going back. Psychologists call this negative reinforcement. The behavior that removes discomfort gets stronger, even when it makes the underlying problem worse.
In my work with clients, I see this cycle constantly. The relief gets shorter each time. The urge to check comes back faster. What started as one question becomes a nightly ritual.
Every time you check, you send yourself a quiet message: I cannot tolerate not knowing this. Your brain listens. It learns that the doubt must be a real threat, because look how urgently you respond to it.
The worry gets louder, not quieter, the more you chase certainty. And AI removes the natural friction that used to slow this down. A friend eventually says you have asked me this five times. A chatbot never does. Nothing interrupts the loop, so the loop runs.
For some people, this pattern is bigger than a habit. Reassurance seeking is one of the most common compulsions in OCD, where an intrusive doubt demands an answer and no answer is ever quite enough. It also shows up in chronic second-guessing, health anxiety, and relationship anxiety.
The content changes. The mechanism does not. A doubt appears, it feels intolerable, and checking buys a moment of peace at the cost of a stronger doubt tomorrow. If that describes your relationship with AI, the tool is not the cause. It is the newest and most frictionless place the old pattern has found to live.
Here is the uncomfortable truth. You do not need one more source. You need practice letting the uncertain feeling be there without doing something about it right away.
Start small. When the urge to ask hits, wait ten minutes. Notice what the doubt feels like in your body. Name it: this is anxiety asking for reassurance, not a question that needs answering. Then let the ten minutes pass without opening the chat.
I often tell clients that the goal is not a better answer. The goal is discovering that you can survive the not knowing. Each time you sit with the doubt instead of feeding it, the anxiety loses a little of its grip. That is the same principle behind exposure-based treatment, and it works.
AI can be genuinely useful for a lot of things. Answering the same worried question for the fifth time is not one of them. If you want a deeper look at where these tools help and where they fall short, we have written about whether AI can replace a therapist.
If you keep reaching for reassurance and it never quite lands, that is something we work on. The therapists at Good Therapy SF help adults in San Francisco who feel stuck in worry, checking, and doubt find steadier ground. Reach out to learn more.
Dr. Tom McDonagh, PsyD, is a licensed clinical psychologist and the founder of Good Therapy SF. He trained at Vanderbilt and completed his postdoctoral work at Kaiser Permanente. His practice focuses on anxiety, OCD, panic, and work stress, with a special focus on San Francisco professionals. Good Therapy SF is located at 870 Market Street, Suite 617, San Francisco, CA.
