tI ran* I sat from me, called my hand and scrolled. “I just wanted to make sure I didn't say the wrong thing,” he explained, referring to his recent differences with his partner. “So I asked ChatGpt what to say.”
He read the messages generated by the chatbot aloud. It was made up of clear, logical – almost composed. It didn't sound like a tran. And it certainly didn't sound like I was in the middle of a complicated, emotional conversation about the future of a long-term relationship. Nor did I mention some of Tran's contribution actions to the tensions in the relationship that Tran and I were discussing anywhere.
Like many others I've seen in treatment recently, Tran had his eyes on AI at the moment of crisis. Under heavy workplace pressure, facing uncertainty in his relationship, he downloaded ChatGpt “just to try” on the phone. What began as curiosity quickly became a daily habit, asking questions, drafting texts, sought relief in his own emotions. The more Tran used it, the more he began to re-estimate himself in social situations and turned to models for guidance before dealing with his colleagues and loved ones. He felt strangely comforted, like “No one knows me better.”
Meanwhile, his partner began to feel like she was talking completely to someone else.
CHATGPT and other generative AI models present attractive accessories or alternatives to traditional therapies. It is often available free of charge 24/7, providing detailed, customized responses in real time. When you're overwhelmed, unable to sleep and desperate to understand a troublesome situation, it's very appealing to enter a few sentences into a chatbot to get back what feels like wise advice.
But as a psychologist, I am increasingly worried about what I see in the clinic. A quiet change in how people handle pain and their reliance on artificial intelligence instead of human connection and therapeutic support.
AI may feel like a lifeline when services are overgrowing. And don't make mistakes, service is growing too much. Globally, one in eight people suffered from mental illness in 2019, and there is a severe shortage of trained mental health professionals. Australia is seeing an increasing shortage of mental health workforce that affects access to trained professionals.
Clinician time is one of the rarest resources in healthcare. It's understandable that people are looking for alternatives (even predictions). Turning to chatbots for emotional support is not without risk, especially when the line between advice, security and emotional dependence is blurred.
Many psychologists, myself included, encourage their clients to build boundaries regarding the use of ChatGPT and similar tools. Its enchanting “always on” availability and friendly tone can reinforce unintentionally useless behavior, especially for people with anxiety, OCD, or trauma-related issues. For example, seeking peace of mind is an important feature of OCD and ChatGpt, and by design, it provides a sense of security of richness. Never ask why you are asking again. There is no challenge to avoid it. “Sit down with this feeling and practice the skills we've been working on.”
Tran is often paraphrased until the model gives him the answer he “feels right.” But this constant tailoring meant he wasn't just looking for clarity. He was outsourced emotional processing. Instead of learning to endure pain and explore nuances, he sought the certainty generated in AI. Over time, it made it difficult for him to trust his own instincts.
Beyond psychological concerns, there are real ethical issues. Information shared with ChatGPT is not protected by the same confidentiality standards as registered AHPRA experts. Openai says that data from users will not be used to train models unless authorised, but the vast amount of fine print in user contracts is often unread. Users may not understand how input can be stored, analyzed and potentially reused.
There is also the risk of harmful or incorrect information. These large language models are automated networking. They predict the next word based on previous patterns: This stochastic process can lead to “hastisation,” self-confidence, and sophisticated answers that are not entirely true.
AI also reflects biases embedded in training data. Research shows that generative models can perpetuate and even expand stereotypes based on gender, race and disability. Human therapists also have clinical skills. When a client's voice is shaking, or they realize their silence may say more than words.
This does not mean that AI has no place. Like many technological advances before that, generative AI remains here. In areas where access to mental health professionals is severely restricted, they may provide useful summaries, psychoeducational content, and even support. However, it should be used with caution and never as an alternative to relational and regulated care.
Tran wasn't wrong to ask for help. His instinct to understand pain and communicate more thoughtfully was logical. However, being very leaning towards AI meant that his skill development had suffered. His partner began to notice a strange separation in his message. “It didn't sound like you,” she told him later. In the end, it wasn't.
She also felt dissatisfied with the lack of accountability in his response to her, which caused more relationship friction and communication problems between them.
When Tran and I worked together in therapy, we explored what led him to seek certainty in a chatbot. We unraveled his fear of disappointing others, his discomfort at emotional conflict, and his belief that perfect words might prevent pain. Over time he began writing his reaction.
A good treatment is related. It thrives with imperfection, nuance and slow discovery. It includes the types of discomfort that lead to pattern recognition, accountability, and lasting change. Therapist doesn't just answer. They ask, they challenge. They hold space, provide reflections, and provide uncomfortable mirrors while walking with you.
In the case of Tran, shifts were more than just limiting the use of ChatGpt. It was to regain his own voice. In the end, he didn't need a full response. He had to believe that he could navigate the messiness of life with curiosity, courage and care, rather than a perfect script.
