“She doesn’t judge”: why more and more users are trusting AI

“She doesn’t judge”: why more and more users are trusting AI
Anxiety, insomnia, loneliness: more and more French people are turning to AI psychological support, from ChatGPT to psychological chatbots, to confide in them. But what are we really looking for in these exchanges with a machine that listens without ever judging?

Confiding your anxieties to a machine no longer shocks many people. In just a few years, AI has established itself as psychological support. Faced with deadlines for consultation, many turn to these contacts who are accessible at any time.

A 2025 Ipsos survey suggests that nearly 39% of French people regularly use artificial intelligence. A Stanford survey reports that about 85% of respondents feel more comfortable confiding in a chatbot than a human therapist. This shift towards digital confidence says a lot about our relationship to stress, doubts and the need to listen. It remains to be understood what is at stake in these exchanges.

Why AI is becoming a daily psychological support

In practice, a large proportion of users turn to ChatGPT or “psychological chatbots” to talk about anxiety attacks, relationship problems, loneliness or insomnia. The appeal is simple to summarize: free, 24-hour availability, immediate responses and a feeling of total anonymity. The smartphone becomes a sort of miniature consulting room, always in your pocket. The distance is as reassuring as the speed of the responses.

For some, AI truly serves as a soothing ear. Kelly, 17, confides for example: “AI is good for resting and stopping being afraid of everything and nothing”,
she explains in a recent study. She asks intimate questions about her health and receives structured answers, with advice for action. This type of exchange gives the impression of regaining a little control, even if the precision of the information takes second place.

From the Pythia of Delphi to chatbots: the return of oracular logic

Several researchers relate these uses to ancient oracles. In ancient Greece, the Pythia of Delphi not only gave prophecies: she helped to understand events and make decisions. Today, the ritual is close: faced with uncertainty, we ask a question, we rely on an entity deemed expert, we receive an interpretation and the anxiety subsides a little.

This function does not only concern AI. On YouTube or Twitch, news “decoders” also play the role of new digital soothsayers by explaining the world to their community. In all cases, appeasement takes precedence over factual truth: what matters is feeling that someone is bringing a little order to the chaos. Conversational artificial intelligence extends this human need for meaning, with a smooth and very available interface.

Advantages and limitations of AI as psychological support

Used well, artificial intelligence serves above all as a guided diary: it helps to create words, classify thoughts, suggest breathing exercises or simple courses of action.

But these systems do not detect a psychiatric disorder, bear no medical liability and may encourage dependence. The psychologists point out that they remain complementary support and that in cases of distress, human face-to-face contact remains central.