Study finds distressed people increasingly turn to AI for emotional support

The rise of AI-mediated support

A new study by Tel Hai College of the Galilee has identified a growing trend in how people manage their mental well-being. Researchers found that a significant portion of the population is now using generative AI tools, such as ChatGPT and Gemini, to seek emotional support and coping advice. This behavior is particularly prevalent among individuals who are currently experiencing high levels of psychological stress.

The research, published in the journal Cyberpsychology, Behavior, and Social Networking, surveyed 584 Israeli adults who had used generative AI at least once in the past year. The data shows that while nearly one in three general AI users sought emotional help, that number jumped significantly for those in distress. Specifically, 47% of participants with severe psychological distress reported consulting with AI for mental health-related assistance.

According to the study, emotional distress is the strongest psychological variable linked to this behavior. While many people use these tools for tasks like writing code or drafting emails, their conversational nature makes them attractive for personal use. Dr. Dorit Hadar Shoval, who led the study, noted that these tools offer immediate and nonjudgmental interactions that many people find appealing during difficult times.

Why users are turning to chatbots

Several factors contribute to why people choose to talk to an algorithm rather than a human professional. One major factor is the accessibility of the technology, which is available at any hour of the day. Unlike human friends, AI systems can appear to remember personal details and rarely challenge the user, which can create a sense of constant availability and approval.

The study suggests that social taboos and a lack of trust in people also play a role in this shift. Some individuals may find it easier to admit to a chatbot about sensitive topics, such as past abuse, because the machine does not judge them. This allows users to express themselves in ways they might feel uncomfortable doing in a traditional face-to-face setting with a counselor.

Furthermore, there is a global shortage of accessible public mental-health professionals. Because human help can be difficult to find or obtain, people are turning to AI as a form of 'self-medicine.' This is similar to taking vitamins without a doctor's advice; it is a way for people to manage their own symptoms when professional resources are unavailable or intimidating.

Risks of digital self-medicine

While AI can provide a sense of comfort, the researchers highlight substantial risks associated with its use for mental health. Generative AI systems are not designed to function as clinical providers and lack genuine empathy or an understanding of human emotional states. A chatbot can sound confident and attentive, but it cannot actually feel the emotions of the person it is interacting with.

One of the primary concerns is the lack of clinical boundaries and crisis-intervention capabilities. Because these tools carry no clinical mandate, they may not have the necessary safeguards to handle a user in an acute mental health crisis. The study warns that while a chatbot might agree with a user or offer approval, it lacks the ability to diagnose danger or provide the nuanced care required for complicated problems.

There is also the possibility of a paradoxical dynamic where AI might worsen the isolation it aims to soothe. The researchers describe this as a loop where actions taken to solve a problem might actually feed back into making that problem harder to reach. Because the study does not yet know if these interactions help or hinder coping, they emphasize the need for more research into whether AI serves as a complement to or a replacement for professional care.

Demographics and future research

The study found that the use of AI for mental health is not distributed evenly across the population. Younger participants and women were found to be much more likely to report using these tools for emotional support than older people and men. This suggests that different demographic groups may have different levels of trust in or reliance on digital tools for personal well-being.

Researchers also noted that people with higher levels of attachment anxiety were more likely to seek help from AI, though distress was the more significant predictor overall. Dr. Hadar Shoval noted that the study focused on emotional stress rather than clinical mental illness. However, she stressed that the widespread use of these tools by distressed individuals demands closer attention from policymakers and clinicians.

Moving forward, the research team plans to investigate whether psychological distress drives people toward AI or if infrequent reliance on AI contributes to increased distress. They also aim to study the quality of crisis responses and the safety of these interactions. Understanding these digital help-seeking behaviors is becoming a vital part of modern mental health care as technology becomes more integrated into daily life.

Clinical implications for the future

The findings suggest that clinicians and researchers need to recognize AI as a part of modern self-care. Because people are increasingly using AI for 'self-medicine,' clinicians need to be aware of these digital behaviors. Understanding how patients use AI can help providers offer better, more informed care that acknowledges the tools people are already using to cope with their stress.

For policymakers, the findings highlight the need for clearer guidelines on the use of generative AI for mental health purposes. Since these tools are not designed as clinical providers, there is a risk that they could provide incorrect information or fail to intervene during a crisis. Establishing safety standards and crisis-response protocols is essential as these tools become more common in the public sphere.

The researchers emphasize that the goal is not to stop people from using AI, but to understand its role in the mental health ecosystem. They want to know if AI serves as a complement to professional support or if it replaces it in some cases. By studying these patterns, the community can work toward a future where technology safely supports human well-being without compromising safety or clinical integrity.