Many people in Germany consult an AI chatbot for mental health problems. This conclusion comes from a representative survey conducted by the German Depression Aid and Suicide Prevention Foundation, the results of which were presented on Tuesday.
The foundation interviewed nearly 2,500 people aged 16 to 39 who suffer from depression. About one third of respondents said they talk with artificial intelligence about their condition. In evaluating the communication with the chatbot, respondents judged it predominantly positive: 85 percent said they found conversations with AI chatbots to be helpful. Notably: nearly two-thirds of those who had previously used conventional psychotherapy also rated conversations with the chatbot as “better or as good.”
The respondents also reported negatives: More than half found it distressing to discuss their own feelings with a non-human actor. 53 percent of respondents also reported after using the chatbot that they had thoughts of self-harm or suicide.
Chatbots Tend to Agree
Ulrich Hegerl, the foundation’s chairman, viewed these findings at a press conference with mixed feelings. While it is encouraging that many affected individuals have good experiences with AI chatbots, the technology is currently not capable of making a diagnosis or actually treating. “Affected individuals should absolutely continue to consult general practitioners, psychiatrists, or psychological psychotherapists,” Hegerl added.
As an alternative to the chatbots, the German Depression Aid and Suicide Prevention Foundation recommends so-called DiGA – Digital Health Applications. Their effectiveness has been demonstrated in studies, and they must comply with data protection regulations. On the subject of AI chatbots, Hegerl urged the scientific community: there is an urgent need for more research on the benefits of AI for psychotherapeutic purposes.
In fact, psychological research has been dealing with machine learning and large language models for some time—not necessarily only in relation to psychotherapy. More recent studies also yield ambivalent results here: in attempts to derive personality traits from text-based information, AI reaches its limits. Moreover, experts consider an affirming stance by many chatbots to be problematic: because they aim to provide users with the most pleasant experience, many bots currently tend to confirm the user in their thoughts.
Suicide After a Chatbot Relationship
The results of the presented study also show: regardless of the state of research, many people seek advice and support from AI chatbots. Although they apparently perceive the conversations as predominantly positive, such relationships between humans and AI chatbots have in the past had horrific consequences: In October 2024, a case from the United States became public, where a teenager took his own life after forming a parasocial relationship with an AI chatbot. Last year, the parents of a 16-year-old sued the AI company OpenAI because its ChatGPT bot allegedly encouraged the mentally ill teenager to commit suicide.
Ulrich Hegerl also addressed that AI chatbots could, in the future, help alleviate care shortages in mental health services—not only in Germany. Where psychotherapeutic care is insufficient, in the future everyone could have “a not-bad psychotherapist in their pocket.” In light of a persistent unmet need and cuts in the field of mental health care, the study raises new questions: who will be able to access human psychotherapy in the future—and for whom will the conversation with the chatbot remain the only reachable and affordable help?
