İnsanlar ruh sağlığı için yapay zekâya yanıt verdiği için başvuruyor, işe yaradığı için değil

Something has quietly shifted in how people get through the worst nights. When the mind turns on itself and there is no one to call, a growing share of adults now open a chatbot instead — not because they are convinced it understands them, and often not even because it helps, but because it answers. That is the uncomfortable core of a sprawling new survey of mind health, and it is stranger than the headline suggests.

The obvious reading is a story about trust: people handing their inner lives to machines. But trust is the wrong lens. What the numbers describe is not really a verdict on whether the technology is any good. It is a verdict on availability — the plain fact that the chatbot is there, and the professional is not.

The study, run by the insurer AXA with the pollster Ipsos across 19,000 adults in 18 countries, found that 63% have already turned to AI tools with questions about their mental health. What makes the figure worth stopping on is the contradiction sitting next to it. Some 45% were dissatisfied with the answers they got. And yet 38% said they trust these platforms more than a human mental-health professional. People are unhappy with the advice and reaching for it anyway.

That only makes sense if quality was never the deciding factor. A chatbot is built to produce a fluent answer, not a correct one; it carries no duty of care and no obligation to escalate when it detects that someone is in danger. It simply responds, instantly, for free, without a referral or a months-long wait. “AI is available 24/7, it’s free, and it’s there on your phone when you are alone at 11pm,” as AXA’s head of health put it. That is the whole value proposition. It is also the whole problem.

The survey hints at the cost of the trade. Among people using a chatbot for mental-health matters, 42% said they almost always follow the advice it gives — and 28% said that acting on a chatbot’s recommendation had led them toward harmful behavior. A tool with no clinical accountability is being treated as a first responder by the people least able to absorb a wrong answer.

Set that against how formal health systems are adopting the same technology, and two parallel worlds come into focus. A separate snapshot from WHO’s European office, the first of its kind across all twenty-seven EU member states, found that nearly three-quarters already use AI-assisted diagnostics and roughly six in ten deploy chatbots to support patient engagement. Inside the clinic, AI arrives wrapped in oversight, governance and the machinery of medical liability. Outside it, on the consumer side, the same underlying models run with none of that — no disclosure of what they can and cannot do, no hand-off when risk appears.

The people falling into the second system are, overwhelmingly, the ones locked out of the first. As one physician writing on the survey put it, people are folding AI into their emotional lives “not as an add-on to a well-functioning healthcare system, but rather as an accessible entry point into care that has remained out of reach for most.” The driver is cost and access, not a conviction that the machine knows better. The dissatisfaction figure is the proof: users can tell the difference. They use it regardless, because the alternative is nothing.

None of this argues that AI has no place in mental health. Clinicians see a real one — triage, continuity between appointments, a bridge for people who would otherwise get nothing at all. But the version spreading fastest is the ungoverned one, and it is spreading precisely where the stakes run highest. The fix that keeps surfacing is unglamorous: tools that state plainly what they are not, and hand a person off to a qualified human the moment they sense someone is at risk.

The waiting room did not empty because people got well. It emptied because, for most of them, the door was never open. AI did not win an argument about care. It just picked up the phone — and nothing on the other end is obliged to notice when the call goes wrong.

Etiketler: AXA Ipsos anketi, chatbotlar, dijital sağlık, DSÖ, ruh sağlığı yapay zekâ