Looping when another answer does not help

Markéta received a message from her partner without the usual greeting. She asks whether it means his feelings have cooled. The chatbot offers several possibilities. Markéta feels relief, then notices the full stop at the end and asks again. The responses may remain cautious, yet the conversation supports a repetition that exhausts her.

We call this pattern a loop, or looping, here. Markéta looks for an answer that will finally remove uncertainty. She gets it for a moment, then another detail emerges that needs interpreting. The same message can be analysed again and again without Markéta knowing any more about her partner’s intention than at the start. Answers accumulate; understanding does not.

Am I missing new information, or do I need to feel certain again? This question helps us notice the difference. Repeated reassurance seeking can be part of obsessive-compulsive disorder, or OCD. But an ordinary request for support, a long conversation or a question asked twice does not establish this condition.^[1] We use ‘loop’ to describe the course of a conversation, not to assign a diagnosis.

Asking follow-up questions is different

When you do not understand an explanation, another question may be exactly what you need. ‘Not yet; show me with another example’ returns to the same subject to build understanding. You may likewise want to speak about a painful experience again. Repeating a subject should not itself be a reason for the assistant to brush you off.

Watch what another response gives you. Did it explain an unclear word, help find your own wording or open another part of the experience? Or do you merely hear a reassuring conclusion again, whose certainty soon stops being enough? On screen, the two conversations may look similar. They differ in why they continue and what changes between responses.

Similar repetition can arise when checking AI itself: ‘Is it still the same agent?’ ‘Did it really understand me?’ ‘Did the checker really find no error?’ If you keep working with the same information, another assessment may add only more temporary reassurance. For such a moment, it helps to have a previously chosen option to return to: dinner, an arranged call or another activity you wanted to do. You do not need one final soothing sentence or the assistant’s permission to stop.

‘No new information has arrived now. I don’t have to get the last answer to be able to stop. I’ll return to the question when I have a reason to address it differently.’

This approach is not an instruction to ignore continuing danger. Repeated violence need not bring new information, yet you still need protection. Nor is a new health symptom or genuinely not understanding an answer automatically a loop. Stopping the chat does not mean stopping the search for help.

If checking exhausts you, affects your functioning or you cannot limit it, discuss it with a professional. In OCD treatment, follow the agreed plan; do not devise your own exposure procedure with a chatbot. Someone close can remain supportive even while refusing to decide for a sixth time what the same full stop in a message means.

When using it stops being your choice

Concern about dependence may begin with ‘I’m there all the time’. Duration alone does not tell us what is happening. Someone writes for a long time because an assistant helps them manage a demanding task. Someone else cannot tear themselves away from a conversation that no longer helps and repeatedly displaces sleep or something they care about. Control over use and its effects are what matter.

When you have this concern, try describing specifically when you wanted to stop, what brought you back and what you did not do because of it. A one-off evening and a repeated pattern you cannot change despite its consequences need different solutions. Liking an assistant, writing to it daily or missing its familiar way of responding does not itself tell us whether you have lost the ability to choose.

Help can begin with this pattern without first deciding on a diagnostic name. If you set boundaries but repeatedly cannot maintain them and use is harming you, discuss the specific difficulties with a professional. The goal may be to regain sleep, concentration or the freedom to stop a conversation while keeping features that truly serve you. If AI is your only available support, further real support must be sought alongside it. Simply removing an application does not itself create a person you can confide in.