When AI Reassurance Makes Anxiety Stronger

The Atlantic's How to Touch Grass podcast recently explored a question that deserves more attention: can AI chatbots make anxiety worse even when they feel helpful in the moment?

The episode features Stanford psychiatrist Elias Aboujaoude and psychologist Ashleigh Golden. Their concern is not simply that a chatbot might produce incorrect information. It is that the design of the interaction can reward anxious behaviour. The tool is always available, sounds confident and often responds in a warm, agreeable tone. That makes reassurance immediate and very easy to repeat.

Relief is not always resolution

Imagine that you are worried about a health symptom, a work decision or an upcoming journey. You ask a chatbot for reassurance and feel calmer. A few minutes later, uncertainty returns. You ask a slightly different question, request another comparison or refine the scenario again.

Each answer may reduce anxiety briefly. The repeated checking can still teach you that uncertainty is dangerous and that you need the chatbot to feel safe. The episode explains that this is different from learning that anxiety can rise and fall without a compulsion.

This distinction matters. A tool can answer the question you asked while missing what you actually need. A trusted friend, coach or therapist may notice that you are no longer gathering useful information. You are looking for a guarantee that no tool can honestly provide.

Chatbots remove natural stopping points

A search page eventually runs out of useful results. A human conversation has limits. A chatbot can keep offering another angle, another optimisation and another prompt. It rarely says that you have enough information and should now stop.

That endless helpfulness can be a problem for people who already overthink. The Atlantic episode gives examples of planning and work tasks expanding because AI makes it possible to keep refining. More options do not always create more clarity. Sometimes they create a larger space in which doubt can continue.

The issue also reaches the workplace. Many employees are now expected to use AI. If every document can be revised again and every plan can be optimised further, the tool that promised speed can increase perfectionism, working hours and mental load.

How I would set healthier boundaries

I do not think the answer is to avoid AI completely. I use ChatGPT in my own work and find it valuable. The answer is to decide what role the tool should play before an anxious moment decides for you.

Start by defining the task. Are you collecting facts, preparing questions, comparing a small number of options or trying to calm a feeling? AI can help with the first three. When the real goal is repeated reassurance, continuing the conversation may not help.

Set a stopping rule. Give yourself one research session, three options or 20 minutes. Write down the decision criteria before asking the tool. Once you have enough information, close the conversation and do something in the real world.

Keep human circuit breakers. Discuss a medical concern with a qualified professional. Bring a difficult decision to someone who knows you. If anxiety is persistent or interfering with daily life, seek support from a mental-health professional rather than treating a chatbot as therapy.

Finally, practise leaving some uncertainty unresolved. A responsible decision does not require perfect certainty. It requires enough information, a clear value and a next step you can review.

My take

The most important question is not whether AI feels supportive. It is whether the interaction leaves you more capable after you close the screen. Did it help you act, learn or communicate? Or did it make you more dependent on another answer?

For people navigating career change, this is especially relevant. There will never be a perfect forecast of which job, company or business idea will work. ChatGPT can help you prepare and test assumptions. It cannot remove the uncertainty that comes with building a new chapter.

I recommend listening to and reading The Atlantic's original episode: How AI Could Make People More Anxious.