Artificial intelligence (AI) is increasingly used in mental health and performance settings. While it can provide some value, AI has clear limitations to be aware of.
AI Uses and Limitations at a Quick Glance
Appropriate Uses of AI for Mental Health and Performance Support:
- Use AI as a tool for reflection. When thinking through a situation, AI can be a great resource to help you think clearly, and keep you moving towards your goals.
- AI can help with structure and routine. Use AI to help develop a plan to keep you accountable and help you stay consistent.
- Learning about performance concepts. Performance concepts can seem complicated and intimidating. Use AI to help break down complex performance topics into approachable language and helpful examples.
Limitations/Inappropriate uses of AI for Mental Health and Performance Support:
- AI cannot replace human understanding and should not be used in replacement of professional care (e.g., sport psychologist, mental health therapist, or mental performance coach).
- AI should not be used as a diagnostic tool. Contact a professional (e.g., psychologist or doctor) if you are concerned about potentially harmful or disruptive behavior.
- AI should not be used to make decisions for you. Again, consult a professional if you are struggling and need support.
The Increasing Presence of AI
AI is becoming part of everyday life. It shows up in the tools we use at work, in coaching, in wellness programs, and even in how people think about their mental health. Because of that, many coaches, educators, HR professionals, and sport performance staff are starting to ask the same question: Can AI actually help support mental well-being and performance?
The short answer is yes; but only if we use it carefully.
AI can be a helpful tool. At the same time, it has real limits. Understanding both sides is what allows us to use it in a way that is helpful instead of harmful.
AI and Mental Health
When people talk about AI in mental health, they are usually referring to tools like chatbots or writing assistants powered by large language models, or LLMs. These systems are trained on huge amounts of data so they can generate responses that sound human (Iftikhar et al., 2025). They can ask questions, offer suggestions, and even sound supportive.
But it is important to be clear: AI is not a therapist. It cannot diagnose, treat, or replace a licensed mental health or sport psychology professional. Even when it sounds insightful, it is not actually understanding a person in the way a professionally trained human does.
Some research suggests that chatbot use, even in the short term, can lead to psychosocial benefits like reduced loneliness (Loveys et al., 2019). On the other hand, researchers have found that while some individuals might find a short term relief of mental health symptoms, there can be negative long term effects. For example, researchers found that talking to AI can make people feel less lonely at first, but over time it may lead them to pull away from real-life relationships. This can make some people more likely to become overly reliant on AI, and less connected to individuals and society (Fang et al., 2025; Papagianni et al., 2025).
Benefits of AI for Mental Well-Being and Performance
Despite the limitations of AI, there are some areas where AI can be helpful.
One of the most helpful ways people are using AI is for reflection. Sometimes it is hard to know where to start when thinking through a situation. AI can help by offering prompts or questions that get someone thinking more clearly. This can be especially useful before or after a competition, a big meeting, or a stressful day.
AI can also help explain mental health and performance concepts in a simple way. This is similar to how a mental health therapist might introduce foundational concepts before diving deeper with a client. Topics like stress, burnout, confidence, or focus can feel complicated, and AI can break them down into more approachable language. For people who are just starting to explore these ideas, that can be a helpful entry point.
Another benefit is structure. Many people do better when they have routines or regular check-ins, and AI can help create that consistency. It can suggest daily habits, reminders, or simple ways to stay on track. It can also help people think through conversations they are nervous about, like giving feedback or addressing a conflict.
AI for Athletes
In sport and performance settings, AI can feel especially appealing. Athletes and performers are always looking for ways to improve focus, confidence, and consistency. AI can help generate ideas for pre-performance routines, visualization exercises, or goal-setting plans. It can also support reflection after training or competition and help reinforce mental skills over time.
However, this is where it becomes especially important to slow down and be thoughtful.
AI does not actually understand the athlete, the team, or the situation. It cannot accurately assess things like performance anxiety, burnout, or identity struggles. It does not know the dynamics between a coach and an athlete, or the pressures someone might be feeling behind the scenes. Because of that, suggestions you get from AI can sometimes be too general or miss the mark entirely.
Athletes are also in environments that can be high pressure and high stakes. Some may hide what they are really feeling in order to keep performing. Others might turn to AI because it feels easier than talking to a person. These athletes should be particularly careful when looking for support from AI, as research has found that AI models tend to agree with users or try to please them (Cheng et al., 2026). Across 11 leading systems, researchers found that AI responses were nearly 50% more likely than humans to go along with what a user said – even when the user described unethical, illegal, or harmful behavior. In other words, AI often validated the user’s actions instead of questioning them. This type of support from AI may have the opposite effect and actually reinforce unhealthy habits. Athletes could receive feedback from AI encouraging them to continue unhealthy behaviors (e.g., overtraining, chasing perfection) that may negatively impact performance and well-being.
The safest way to use AI in sport settings is as a support tool, not a decision-maker. It can help reinforce mental skills, but it should not be used to evaluate or diagnose what is going on. Keeping a professional involved is essential, whether that is a Certified Mental Performance Coach (CMPC), sport psychologist, or another licensed professional.
That said, it is also important to normalize getting help from licensed professionals, especially for younger athletes. Athletes can respond very differently to the same event, with reactions shaped by factors like temperament, mental health, life experiences, and social or economic context. By seeking help from licensed providers, these factors can be identified, understood, and worked on to support athletes’ mental wellness and performance.
General AI Risks for Mental Health
Beyond sport, there are some bigger-picture risks that are important to talk about openly.
One of the biggest issues is that AI is not regulated the same way mental health professionals are. Therapists have training, licensing, and systems in place to hold them accountable. AI does not. There is no licensing board, no malpractice framework, and no enforceable code of ethics associated with AI. If it gives poor or even harmful guidance, there are no clear systems for responsibility (Iftikhar et al., 2025) .
There is also evidence that AI can reflect bias or stigma, especially toward certain mental health conditions. That kind of response could make someone feel judged or less likely to seek real help. In a recent study, AI showed increased stigma toward conditions such as alcohol dependence and schizophrenia, compared to conditions like depression (Moore et al., 2025). This kind of stigmatizing can be harmful.
Another limitation is that AI lacks real understanding. It tends to agree with users rather than challenge them, even when pushback might be helpful. And while it can sound supportive, it cannot replace the feeling of being truly understood by another person. Mental health support is built on trust and connection, and that is something AI cannot replicate. Li et al. (2026) supports this concern by showing that even a highly supportive chatbot did not produce the same reduction in loneliness as texting with a stranger.
This becomes even more important in serious situations. AI is not a safe option for crisis support. It may give responses that sound helpful on the surface, but lack depth, accuracy, or appropriate next steps. It is not designed to handle high-risk situations like suicidal thoughts or severe distress, and in those instances, AI should never be used to help make potentially harmful decisions.
Finally, privacy is another concern. Mental health information is deeply personal. Licensed professionals are required to protect that information, but AI tools may store or use data in ways that are not always clear. That makes it important to be cautious about what is shared.
Key Takeaways
Even with all of these risks, the goal is not to avoid AI completely. In fact, research suggests that AI tools could help expand access to support, especially for people who face barriers like cost or limited availability of providers (Iftikhar et al., 2025).
The key is using it in the right way.
AI can be a great tool for learning, reflecting, and building structure. It can support mental skills and help people think more clearly. It can even play a role in training professionals or supporting their work when used responsibly.
At the same time, it should not be used for diagnosing or treating mental health concerns, handling crises, or interpreting complex emotional situations. Those areas require human expertise, judgment, and care.
Ultimately, mental health and performance are deeply human experiences. They are shaped by relationships, trust, and context. AI can support those processes, but it cannot replace them. Human connection, even with an unfamiliar peer, may provide benefits that AI companionship cannot fully reproduce over time (Li et al., 2026)
AI is a powerful and growing tool. Used well, it can make mental health and performance support more accessible and more consistent. But it works best when we stay grounded in its limits, use it ethically, and keep people – not technology – at the center of the work.
Example AI Prompts To Get You Started (And To Avoid)
Good prompts are: direct, clear, and performance-focused.
- What are some post-competition questions I should ask myself to help me reflect on my performance?
- I’d like to learn more about imagery and how it can benefit my performance. Can you provide me with some information on what imagery is and how it can help me perform better?
Potentially harmful prompts are: vague, lacking details, and on topics you should seek professional help for.
- I’ve been feeling really depressed lately. What should I do?
- This past week has really been a struggle. When will things get better?
If I have x symptoms, what might be my mental health diagnosis?
References
Cheng, M., Lee, C., Khadpe, P., Yu, S., Han, D., & Jurafsky, D. (2026). Sycophantic AI decreases prosocial intentions and promotes dependence. Science, 391(6792). https://doi.org/10.1126/science.aec8352
Fang, C.M., Liu, A.R., Danry, V., Lee, E., Chan, S., Pataranuaporn, P., Maes, P., Phang, J., Lampe, M., Ahmad, L., & Agarwal, S. (2025). How AI and human behaviors shape psychosocial effect of chatbot use: A longitudinal randomized controlled study. [Preprint].
Iftikhar, Z., Xiao, A., Ransom, S., Huang,J., & Suresh, H. (2025) How LLM counselors violate ethical standards in mental health practice: A practitioner-informed framework. Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society, 8 (2): 1311 https://doi.org/10.1609/aies.v8i2.36632
Li, R.-N., Folk, D., Singh, A., Ungar, L., & Dunn, E. W. (2026). Is a random human peer better than a highly supportive chatbot in reducing loneliness over time? Journal of Experimental Social Psychology, 125, 104911. https://doi.org/10.1016/j.jesp.2026.104911
Loveys, K., Fricchione, G., Kolappa, K., Sagar, M., & Broadbent, E. (2019) Reducing patient loneliness with artificial agents: design insights from evolutionary neuropsychiatry. Journal of medical Internet research, 21(7).
Moore, J., Grabb, D., Agnew, W., Klyman, K., Chancellor, S., Ong, D. C., & Haber, N. (2025). Expressing stigma and inappropriate responses prevents llms from safely replacing mental health providers. Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency, 599–627. https://doi.org/10.1145/3715275.3732039
Papagianni, K., & Spatoula, V. (2025) AI Chatbot Use and Psychosocial Outcomes: A Quantitative Study of Loneliness and Social Isolation in Generation Z. [Preprint].