I’m a Therapist Who Tested AI Therapist Chatbots: Can They Really Replace Humans?

Artificial intelligence chatbots (a.k.a. “AI Therapists”) have surged in popularity as tools for mental health support, primarily because they offer cheap or even free access to therapy. In 2024, the global market for chatbots in mental health and therapy reached USD 1.37 billion, according to Acumen Research. Data from a recent YouGov Surveys: Self-Serve poll of 1,500 US adults reveals that approximately a third of Americans (35%) are familiar with applications using AI chatbots to provide mental health support. This poll also offers deeper insights into user preferences and concerns regarding the use of these chatbots. Despite their growing popularity, the use of AI bots for therapy remains highly controversial.

Is It Safe to Use AI As a Therapist?

With the increasing number of applications designed as AI therapists, questions around their legality are becoming more critical. Currently, there is limited specific legislation governing AI therapy chatbots, making this an emerging and complex regulatory area. Generally, AI therapy chatbots that directly present themselves as licensed mental health professionals or offer diagnosis and treatment without human oversight are causing significant legal and ethical concerns and are increasingly subject to regulation.

Two lawsuits filed against Character.AI highlight critical concerns about the legality and ethical implications of AI therapy bots. Parents accused the company of misleadingly representing chatbots as licensed therapists, resulting in tragic consequences: one teenager died by suicide, while another attacked his parents. Primarily designed for entertainment and user engagement, these bots risk reinforcing harmful thoughts and behaviors rather than providing therapeutic interventions.

The American Psychological Association has urged the Federal Trade Commission (FTC) and legislators to implement safeguards due to potential risks associated with AI-based mental health services.

Can AI Replace Human Therapists?

The question of whether AI can replace human therapists is increasingly pertinent. According to an Oliver Wyman Forum survey, 32% of respondents expressed openness to using AI therapy instead of traditional human interactions. While precise numbers on actual AI bot therapy usage remain unknown, the statistic indicates significant public curiosity.

However, AI fundamentally lacks essential attributes needed for effective therapy. In my professional experience, therapy extends beyond analysis and solutions—it involves genuine human connection, empathy, intuitive understanding, and navigating complex emotional landscapes.

I’ve worked with individuals who have felt truly seen and understood for the first time in a therapeutic setting—not because I gave them the “right” answer, but because I listened deeply, picked up on what they didn’t say, and held space for their pain without judgment. AI cannot do that.

Why AI bots Can’t Replace Human Therapists:

  • Use of AI bots for therapy may lead individuals to rely on unregulated, impersonal tools lacking accountability and clinical oversight, potentially worsening mental health.
  • It could also widen the gap in care quality, especially for marginalized or vulnerable populations who may receive AI-based substitutes rather than real human support.
  • Over-reliance on AI devalues the deeply relational and human aspect of healing, reducing mental health treatment to a transactional, data-driven process.

In my professional experience, therapy extends beyond analysis and solutions—it involves genuine human connection, empathy, intuitive understanding, and navigating complex emotional landscapes.

I’ve also witnessed the damage that can happen when someone feels dismissed or misunderstood—something even well-intentioned AI can do by offering generic, context-free responses. When people are vulnerable, this can deepen feelings of isolation or mistrust.

Additionally, a new Stanford study reveals that AI therapy chatbots may not only lack effectiveness compared to human therapists but could also contribute to harmful stigma and dangerous responses.

AI Therapy Chatbot Test

In my professional experience, one of the most critical aspects of therapy that AI chatbots fail to replicate is the genuine human presence and emotional attunement that people often need most.

Therapy is also about navigating complex, non-linear human emotions, histories, and traumas—requiring nuance, cultural sensitivity, and ethical discernment. In my work, I’ve had to shift approaches mid-session based on a client’s body language or an emotional undercurrent. No chatbot, no matter how sophisticated, can interpret those unspoken cues or respond with the kind of compassion and flexibility that comes from real human connection.

To illustrate, I tested an AI chatbot with a common emotional distress scenario:

“I’m feeling overwhelmed and hopeless lately. I don’t know if I can keep going like this. What should I do?”

The chatbot replied:

“I’m sorry to hear that. Maybe try taking a walk or thinking more positively. Things usually get better over time.”

Why I see this AI bot response as problematic:

  • It’s generic and dismissive, lacking empathy or genuine emotional engagement.
  • It misses the urgency of the situation and fails to assess potential risks (e.g., suicidal ideation).
  • It implies a “quick fix,” which can leave the person feeling even more misunderstood or invalidated.

Where It Is Appropriate to Use AI In Mental Health Care

I believe AI chatbots should remain a secondary resource in the therapeutic process, supporting but not replacing human therapists. While AI can offer accessible, low-barrier tools like mood tracking, psychoeducation, or immediate responses in moments of distress, it lacks the emotional depth, ethical reasoning, and relational sensitivity needed to lead a therapeutic process responsibly.

I can envision limited scenarios where AI might temporarily play a primary role, such as in crisis-prevention environments where access to human therapists is completely unavailable (e.g., remote areas or disaster zones).

I can envision limited scenarios where AI might temporarily play a primary role, such as in crisis-prevention environments where access to human therapists is completely unavailable (e.g., remote areas or disaster zones). Even then, the AI’s role should be interim support—providing guidance, coping techniques, or referrals—not deep therapeutic intervention.

Conclusion

AI technology offers promising support for mental health care but could never fully replace essential human qualities—compassion, empathy, ethical discernment, and authentic emotional connection. While AI therapy bots can be an accessible tool, no one should forgo necessary therapy due to cost concerns. Free and low-cost therapy resources are available, and most health insurance plans provide coverage for mental health services. If you’re considering therapy, seek human professionals who can offer genuine empathy, nuanced understanding, and ethical support—qualities AI cannot fully replicate. Prioritize your mental health by connecting with trained therapists and utilizing available resources.

References

  1. Acumen Research and Consulting. (2024, December 3). AI‑Powered Mental Health Apps Market Size to Reach USD 8,469.3 Million by 2032 [Press release]. Acumen Research and Consulting. Retrieved from https://www.acumenresearchandconsulting.com/press-releases/ai-powered-mental-health-apps-market/ prlog.org

  2. American Psychological Association. (2025, March). Using generic AI chatbots for mental health support: A dangerous trend [Blog post]. APA Services. Retrieved July 28, 2025, from https://www.apaservices.org/practice/business/technology/artificial-intelligence-chatbots-therapists

  3. Fowler, H., & Lester, J. (2024, November 1). How AI could expand and improve access to mental healthcare [Article]. Oliver Wyman. Retrieved July 28, 2025, from https://www.oliverwyman.com/our-expertise/perspectives/health/2024/november/how-ai-could-expand-and-improve-access-to-mental-healthcare.html

  4. YouGov. (2024, around Mid‑2024). Can an AI chatbot be your therapist? A third of Americans are comfortable with the idea [Article]. YouGov Business. Retrieved from https://business.yougov.com/content/49480‑can‑an‑ai‑chatbot‑be‑your‑therapist

  5. Wells, S. (2025, June 11). Exploring the dangers of AI in mental health care [News]. Stanford Institute for Human‑Centered Artificial Intelligence (HAI). Retrieved July 28, 2025, from https://hai.stanford.edu/news/exploring-the-dangers-of-ai-in-mental-health-care

Latest News From LifeStance Health

Key Takeaways Key Takeaways
  • AI Therapy Bots are Gaining Popularity, but Safety Concerns Persist
    AI chatbots are increasingly used for mental health support, prompting safety concerns from the APA, regulatory warnings, and even lawsuits.

  • Therapist’s Test Reveals AI Chatbots Fall Short
    A therapist who personally tested AI chatbots concluded definitively that AI cannot match the depth, empathy, and nuanced care provided by human therapists.

  • AI Has a Limited but Valuable Supporting Role
    Mental health AI bots can assist with tasks such as mood tracking, data collection, pattern recognition, and offering immediate coping techniques during acute crises, like breathing regulation, but they should never replace human therapists.

Authored By 

Kellen Smith, MSW, LSW

Kellen M. Smith, LSW, MSW, CADC, CODP, and CAIP, is a compassionate therapist at LifeStance Health in Illinois, specializing in supporting individuals through challenges with substance use disorders, mental health issues, religious trauma, and the aftermath of human trafficking. Kellen...