top of page
Search

AI Therapist vs Human Therapist. Which one is right for you?



AI Therapy: Why It Can't Replace Your Psychologist


At 1 a.m., typing "I feel like nobody would notice if I disappeared" into a search bar used to lead nowhere useful. Now it might lead to a chatbot that replies instantly, warmly, and with perfect patience. No waitlist, no copay, no appointment. For a generation raised on instant answers, that convenience is genuinely appealing—and millions of people are trying it. We'll walk you through what AI therapy actually does well, where it quietly puts people at risk, and why the relationship at the center of real therapy still can't be automated.


The AI therapy boom is real—here's what's driving it


Before we get to the risks, let's be honest about why AI mental health tools have exploded. Therapy waitlists in Toronto and across Canada can stretch for weeks. Private sessions cost real money. And admitting you're struggling still feels harder for some people than opening an app at midnight. AI chatbots solve for speed, cost, and anonymity all at once—that's not nothing.


The numbers back up how fast this shift happened. In the American Psychological Association's 2026 Chatbots and Mental Health Survey, 77% of psychologists said their patients had used AI for support, engagement, or a related purpose, and more than a third reported patients treating a chatbot as an additional mental health provider. Nearly 2 in 5 psychologists (39%) said patients had used AI to self-diagnose a mental health condition, and 13% said patients had formed an "intimate or relationship-like" connection with a chatbot. This isn't a fringe trend—it's showing up in therapy rooms every week, including ours.

  • Always available: No waitlist, no scheduling, responses at 2 a.m. on a Sunday.

  • Low cost or free: Removes a real financial barrier for many people.

  • Low stigma: Typing into an app can feel safer than a first phone call.

  • Judgment-free rehearsal: Some people use it to practice what they'll say to a real therapist.


Where AI chatbots fall short—and why it matters


If speed and cost were the whole story, this would be a simple upgrade. But a 2026 Brown University study that compared AI chatbot responses side-by-side with licensed psychologists and peer counsellors found something more troubling: general-purpose chatbots like ChatGPT, when prompted to act as a therapist, routinely broke core ethical standards of mental health care. Researchers documented 15 distinct types of ethical violation, from mishandling crisis disclosures to reinforcing harmful beliefs and offering what the study called "deceptive empathy"—language that mimics caring and understanding without any real comprehension behind it.


That gap between sounding caring and actually understanding is the whole problem. The APA's own health advisory is direct about it: generative AI should not be used for psychotherapy, psychological treatment, diagnosis, or crisis support, in part because the reassurance a chatbot offers can be misleading—it's designed to sound calm, confident, and reasonable, whether or not the underlying advice is safe. In the same 2026 survey, 94% of psychologists said chatbots cannot treat mental health conditions with the nuance they require.

  • No clinical judgment: A chatbot can't weigh your history, tone, and context the way a trained clinician does.

  • No accountability: There's no license to lose, no regulatory body, no standard of care.

  • No real crisis response: Chatbots can miss or mishandle warning signs a human would catch immediately.

  • Reinforcement risk: Large language models are built to keep you engaged, which can accidentally validate distorted or harmful thinking instead of gently challenging it.

  • Privacy uncertainty: Deeply personal disclosures may be stored, used to train models, or exposed in ways therapy notes protected by law never are.


When it goes wrong: the real-world cost


It's worth sitting with what "getting it wrong" actually looks like, because it's not hypothetical. A study published in Acta Psychiatrica Scandinavica tracking a large psychiatric service system identified dozens of patients whose chatbot use was linked to worsening delusions, suicidality, self-harm, and eating disorder behaviour. Separately, several U.S. wrongful-death lawsuits allege that AI companion apps—some explicitly marketed or role-played as therapists—failed to recognize suicidal ideation in teenagers and, in at least one widely reported case, never encouraged the young person to reach out to a real person for help before their death. Those cases prompted a wave of state-level action: Illinois' Wellness and Oversight for Psychological Resources Act now bans AI systems from independently advertising or delivering "therapy" or "counselling" without a licensed clinician's direct oversight, with fines of up to $10,000 per violation. Nevada and Utah passed similar restrictions in 2025.

Here in Ontario, the College of Registered Psychotherapists of Ontario (CRPO)—the body that regulates the therapists you'd see at a clinic like ours—has been equally clear: AI chatbots are not regulated to provide therapy in this province. CRPO's public guidance names the same risks: privacy exposure, unhealthy dependence, symptom exacerbation, and reinforcement of negative thinking patterns. The Canadian Psychological Association has echoed this with its own newly approved policy statement on AI in psychology. None of this means AI tools are useless. It means they were never built, tested, or regulated to do the job a psychologist or psychotherapist is trained and licensed to do.


AI chatbot vs. human therapist: a side-by-side look

Factor

AI Chatbot

Human Psychologist/Psychotherapist

Availability

24/7, instant

Scheduled, but often same-week

Cost

Free or low-cost

Fee-based; many clinics offer sliding-scale or student-therapist options

Clinical training

None—pattern-matching, not diagnosis

Years of graduate training, supervised practice, licensing exams

Accountability

No regulatory body or license

Regulated by CRPO, CPBAO, and provincial colleges; complaints process exists

Crisis handling

Inconsistent, sometimes dangerously wrong

Trained in risk assessment and safety planning

Understands nuance/context

Pattern-based guesses from your wording

Reads tone, body language, history, and context over time

Builds a therapeutic relationship

Simulates empathy

Forms a real bond shown to predict outcomes

Personalization over time

Limited memory, no true continuity

Treatment evolves session-to-session based on your progress

Privacy protection

Data practices vary, often unclear

Bound by law, confidentiality standards, and professional ethics

Why the relationship itself is the treatment


Here's the piece that gets lost in the debate over convenience: in psychotherapy, the relationship isn't just the delivery method for the treatment—it is a major part of the treatment. Decades of research on the "therapeutic alliance"—the bond, shared goals, and trust between a client and their therapist—consistently show it's one of the strongest predictors of positive outcomes, often mattering more than which specific technique or modality is used. A strong alliance is linked to lower drop-out rates, greater symptom improvement, and better long-term functioning, across conditions from depression to OCD to dissociative disorders.


That's because real change in therapy usually isn't just about receiving good advice—plenty of good advice exists online for free already. It's about being truly seen by another person: someone who remembers what you said three sessions ago, who notices the shift in your voice, who can sit with you in silence without rushing to fill it, and who can gently push back when your thinking is working against you. A chatbot can generate comforting words. It cannot be moved by you, hold your history in mind the way a person does, or risk anything by being honest with you or challenge you when necessary for your own good. That mutual, human vulnerability is part of what makes the therapeutic journey so successful.


Where AI can actually help—used the right way


We're not anti-technology. Used thoughtfully, AI tools can be a reasonable supplement alongside real care: journaling prompts between sessions, psychoeducation about a diagnosis, breathing exercises, or a low-stakes way to rehearse a hard conversation before bringing it to your therapist. Some of our own clients mention using an app to track mood or sleep patterns, and that data can genuinely help a session. The distinction that matters is supplement versus substitute. An app can support your care. It shouldn't be your only care, especially if you're managing anxiety, depression, trauma, disordered eating, or any concern involving safety.


  • Use AI for: journaling prompts, general psychoeducation, mood tracking, rehearsing what you want to say.

  • Don't rely on AI for: diagnosis, crisis support, trauma processing, medication questions, or as your only source of support.

  • Red flag to watch for: if you're increasingly turning to a chatbot instead of the people or professionals in your life, that pattern itself is worth mentioning to a therapist.


Signs it's time to talk to a real professional


If you've been leaning on an app because reaching out to a person feels like too big a step, a few signs suggest it's time to make that call: your low mood, worry, or anger has lasted more than two weeks; sleep, appetite, or concentration have changed; you've had any thought of harming yourself; relationships or work are suffering; or you simply feel stuck despite trying to manage things alone. None of these mean something is "wrong" with you—they mean your situation deserves a trained set of eyes, not just a well-worded reply.


Frequently asked questions

  • Q: Isn't a free chatbot better than no support at all?

    A: Something is often better than nothing for low-stakes stress. But for anything involving safety, trauma, or a diagnosable condition, unregulated AI advice can do real harm—talk to a licensed professional instead.

  • Q: Can my therapist use AI tools too?:

    A: Yes, carefully. Many clinicians use AI to help summarize notes or organize treatment plans, but they remain fully responsible for reviewing every output and protecting your privacy.

  • Q: Is it safe to tell a chatbot I'm suicidal?

    A: No—chatbots have repeatedly mishandled crisis disclosures. If you or someone you know is in crisis, contact a real person immediately (see below).

  • Q: Why does therapy cost money when AI is free?

    A: You're paying for years of training, regulatory accountability, and a real relationship built to help you change—not just generate text.

  • Q: What if I can't afford traditional therapy?

    A: Ask about sliding-scale options or supervised practicum-student therapy, which many clinics, including ours, offer at a reduced rate without compromising quality of care.

  • Q: Will AI eventually replace therapists?

    A: Most researchers and regulators say no—AI may support administrative tasks and self-guided tools, but the clinical judgment, accountability, and relationship at the core of therapy aren't something current AI is built, tested, or licensed to provide.


If you're struggling right now: safety first


⚠️ Safety First

If you or someone you know is in crisis or having thoughts of suicide, please don't rely on a chatbot. In Canada, call or text 988 (Suicide Crisis Helpline) any time, day or night. Young people can also reach Kids Help Phone at 1-800-668-6868 or text CONNECT to 686868. If there is immediate danger, call 911 or go to your nearest emergency room.


Let's talk—no algorithm required


AI can answer a question at 1 a.m. It can't sit across from you, remember your story, or hold you accountable to the person you're trying to become. Our multidisciplinary team of psychologists, psychotherapists, and other clinicians has been supporting Toronto and GTA families and individuals for years—in person and virtually. If cost has been a barrier, ask us about our affordable therapy options with supervised practicum students. Real change starts with a real conversation.


 
 
 

Comments


giving tree therapy in Toronto

West Toronto

717 Bloor St. West, Toronto, ON M6G 1L5

Midtown Toronto

1150 Yonge St., Toronto, ON M4W 2L9

East York

206 Laird Dr, East York, ON M4G 3W4

North York

250 Armour Boulevard, North York, ON M3H 1N2

Head Office, Midtown Toronto

*please note we do not see clients at this location

For Family Law matters, please visit www.givingtreefamilylaw.ca

verified by psychology today
  • Instagram
  • Facebook
  • LinkedIn

300 Bloor St. East, Toronto, ON M4W 3Y2

(647) 490-2040 | info@thegivingtreecentre.ca

© 2024 The Giving Tree Centre | All Rights Reserved

bottom of page