05/05/2026
When AI replaces therapy: What are we gaining and losing
Lately, I've noticed a shift in how people seek support. Not just clients, but family, friends, and strangers online often say: "I don’t need to see anyone, AI figured it out for me," or "AI gave me therapy techniques and I didn’t need to pay."
The rise of AI in mental health isn't random; it fills gaps in our current health system. Research shows barriers like cost, stigma, long wait times, and accessibility prevent many from seeking professional help in Australia [Kavanagh et al., 2023].
As a provisional psychologist, I see AI's potential, but also recognize it's not true therapy. The difference matters more than people realize.
What we are gaining:
AI isn't inherently harmful. Used thoughtfully, it can support mental health in meaningful ways:
1. Reflection and processing emotions: AI helps organize thoughts, feelings, behaviors, and identify patterns, drawing from therapies like Cognitive Behavioral Therapy (CBT).
2. Psychoeducation: Accurate AI can teach about anxiety, depression, attachment styles, and coping strategies accessibly, especially for those previously without this knowledge.
3. Between-session support: For those in therapy, AI can reinforce skills, journaling, or reflection between sessions.
4. Lowering the barrier to help-seeking: AI can be a first step, helping individuals realize they might need more support.
What we are losing:
The concern isn't AI use, but its replacement of professional mental health support, which can cause problems:
1. AI cannot truly assess risk: AI lacks accountability and duty of care. It cannot intervene in crises like suicidal thoughts, self-harm, or severe trauma, unlike trained professionals.
2. It can reinforce but not challenge: A good therapist validates but also challenges unhelpful thinking or behaviors. AI can be overly agreeable, reinforcing patterns unknowingly.
3. Lack of understanding: Therapy involves understanding and articulating experiences. AI offers quick answers without fostering deep comprehension, which isn't helpful long-term.
4. No therapeutic relationship: The therapeutic alliance, the client-therapist relationship, is a strong predictor of outcomes [Flückiger et al., 2020]. Being seen, understood, and connected to another human cannot be replicated by AI.
5. Potential for misinformation or overconfidence: AI can present wrong information confidently or miss crucial context. Mental health treatment needs to be tailored, not one-size-fits-all.
So… what now?
AI will likely remain part of our mental health landscape. The goal is intentional, thoughtful use, not rejection.
Think of AI as:
• A tool, not a therapist
• A starting point, not the whole journey
• A supplement, not a substitute
Ultimately, therapy is about human connection and individualized support, something only a human can provide.
Sarah Leung
Provisional Psychologist - Integral Psychology