“Your therapist doesn’t know you as well as I do.”
Truly yours…AI
Mental Health in the Age of AI
Why Human Care Still Matters
By Ana G. Perez-Barrios, MA, LPC, PYSA Clinical Director| FallUp Wellness
Since 1992, October 10 has been observed as World Mental Health Day, an initiative of the World Federation for Mental Health supported by the World Health Organization. It invites us to reflect on our profession and how we can make mental health a priority [1]
And what could be more relevant in these times than reflecting on the role AI plays in today’s world, especially how its tentacles, for better or worse, are reaching into our field, and how people are increasingly turning to platforms like ChatGPT to “help themselves” with mental health concerns?
The American Psychological Association advises caution about using general-purpose generative AI for mental health care. A response can sound warm and convincing while missing important context or offering unreliable guidance [2]. Feeling understood by a tool and receiving appropriate clinical care are different experiences.
When AI enters the therapy room
I was discussing dating with a client, looking closely at the choices she had been making and exploring alternatives that might interrupt a cycle that left her feeling stuck and burned out. I suggested an approach that was unfamiliar to her and somewhat counterintuitive, to try something different from what had not been working: “Ask what she really wanted from him, without playing the hide and seek dating game”.
She told me she had already asked ChatGPT for advice and that it had recommended what I could consider a simplistic, social media, more radical, “don’t show him you are despair kind of answer”.
And then, when she shared my suggestion with the chatbot, she reported that it responded: “Your therapist doesn’t know you as well as I do.”
Was AI throwing down the gauntlet in a duel for my client’s trust or was this simply a challenging response I needed to address thoughtfully, without discrediting her efforts to explore options for her troubles?
I chose the latter and said: “That could actually be true. AI knows your patterns and how you tend to respond in these situations based on the information it has access to. And yet, isn’t insanity hoping for different results when you keep doing the same thing? Let’s try something different, outside your comfort zone, and see what happens.”
AI as our therapy assistant… always under supervision.
That exchange stayed with me. A chatbot’s claim to know someone better than their therapist is not evidence that it does. At the same time, a client may genuinely feel understood by it, and that experience deserves to be explored. I wanted to understand what resonated with her, explain my reasoning, and leave room for us to evaluate the suggestion together.
Since we are all tempted to use AI as our right-hand assistant for almost anything now, it is important to consider ways to guide our clients to get the best out of it and bring those interactions into therapy for further exploration and ethical reflection. APA encourages open discussion between clients and clinicians about AI use [3]. We could suggest that clients who use AI for emotional support consider these questions:
Does this interaction help me take a useful step in my life?
Am I returning repeatedly for reassurance without feeling more able to cope?
Am I sharing information I would want to keep private?
Is this tool supporting connection with people, or becoming my main source of support?
I would love to hear your suggestions in the comments. What other questions would you, as a therapist, add to this list?
Our responsibilities as providers in the age of AI
As mental health professionals, we need to make room for honest conversations about AI. If someone finds a chatbot comforting, curiosity is more useful than shame: What are they seeking? How does it affect their sleep, relationships, choices, or willingness to seek help?
A therapist also has professional responsibilities: assessing safety, protecting confidentiality, explaining treatment, and coordinating care when appropriate. These responsibilities matter particularly when distress is complex or difficult to describe.
We also need to examine our own use of technology. AI may assist with administrative tasks or educational materials, but clinicians must review its output, protect sensitive information, and explain relevant uses to clients. The World Health Organization highlights risks including inaccurate information, bias, privacy threats, and overreliance on automated recommendations.[4]
Final Disclosure
And in case you’re wondering: yes, I used AI to help me write this article—as an assistant, under my careful supervision. The experiences, reflections, and final editorial decisions are my own.
References
1. World Health Organization. World Mental Health Day. https://www.who.int/campaigns/world-mental-health-day
2. American Psychological Association. (2025). Health advisory: Use of generative AI chatbots and wellness applications for mental health. https://www.apa.org/topics/artificial-intelligence-machine-learning/health-advisory-chatbots-wellness-apps
3. American Psychological Association. Discussing AI use in therapy. https://www.apa.org/topics/artificial-intelligence-machine-learning/discussing-ai-use-therapy
4. World Health Organization. (2024). WHO releases AI ethics and governance guidance for large multi-modal models. https://www.who.int/news/item/18-01-2024-who-releases-ai-ethics-and-governance-guidance-for-large-multi-modal-models