Imagine having a difficult day and needing someone to talk to. Perhaps you’re overwhelmed by responsibilities, struggling with loneliness, or simply trying to organize thoughts that feel impossible to untangle. Rather than calling a friend or scheduling an appointment, you open an artificial intelligence chatbot and begin typing. Within seconds, a response appears. It doesn’t interrupt, become impatient, or require an appointment. For some people, that accessibility is comforting. For others, it raises an important question: Are we discovering a useful new tool for emotional well-being, or expecting technology to provide something it was never designed to replace?
The use of AI for emotional support is no longer unusual. According to the American Psychological Association’s 2026 Chatbots and Mental Health Survey, 77% of more than 1,200 licensed psychologists surveyed reported having patients who discussed using artificial intelligence for support, companionship, or other purposes. Approximately 33% said patients had used AI to assist with therapy or treatment, while 34% reported patients using it for reminders, affirmations, or self-discipline. These percentages describe psychologists’ reports of their patients’ experiences, not the percentage of Americans using AI for mental health. Nevertheless, they demonstrate that artificial intelligence has become part of conversations taking place in professional therapy offices.
There are understandable reasons people might find AI helpful. A chatbot can assist someone in organizing their thoughts before a therapy appointment, explaining unfamiliar psychological terminology, or practicing how to approach a difficult conversation. It may also help users work through journaling prompts, create manageable routines, or review coping exercises previously recommended by a qualified professional. Unlike traditional appointments, these tools can be available at virtually any hour. That accessibility may be particularly appealing to people who face financial barriers, limited local mental health services, or discomfort discussing personal concerns. Research into specialized mental health chatbots has also produced encouraging results in some controlled studies. However, promising findings for a particular program should not be assumed to apply to every general-purpose chatbot.
The concerns become more serious when users begin treating AI responses as professional psychological assessments. Chatbots can generate convincing explanations that are incomplete, inaccurate, or poorly suited to someone’s circumstances. They may also respond too agreeably, reinforcing harmful assumptions instead of challenging them. In the APA’s 2026 survey, 36% of psychologists whose patients had developed relationships with chatbots reported noticing some degree of dependency, while 15% reported distorted thinking or delusions associated with chatbot interactions. These were clinicians’ observations, not proof that AI caused those outcomes. Still, the findings highlight why artificial intelligence should not be relied upon to diagnose mental health conditions, manage emergencies, or replace the judgment of trained clinicians.
Privacy and emotional attachment introduce additional complications. People sometimes share deeply personal information with chatbots because the conversation feels private and judgment-free. Yet information entered into an AI service may be stored, reviewed, or processed according to that company’s policies. Users should understand those practices before sharing sensitive details. There is also a meaningful distinction between finding comfort in a conversation and believing the technology understands or cares in the same way another person does. AI can produce language that sounds compassionate, but it does not possess a therapist’s professional accountability or the mutual responsibilities of a human relationship. For someone already experiencing isolation, an always-available chatbot might offer temporary relief while also making it easier to avoid difficult but important human connections.
Ultimately, the question may not be whether AI belongs anywhere near mental health, but how it can be used responsibly. Artificial intelligence can serve as a writing partner, organizational assistant, educational resource, or supplementary tool for practicing healthy coping strategies. It may help someone find the words to ask for support when speaking to another person feels difficult. But its usefulness depends on recognizing its limitations, questioning its answers, protecting personal information, and maintaining access to qualified care when needed. Technology does not have to replace human connection to make a positive difference. Sometimes its greatest contribution is helping someone take the next step toward that connection.
