Marianne Brandon Ph.D.
The Future of Intimacy
PSYCH CAREERS
Personal Perspective: Facing industry shifts is the only way to remain relevant.
KEY POINTS
- An APA survey showed almost 38% of the population studied found LLM’s more helpful than traditional therapy.
- AI outperformed human group therapy in a major RCT. The profession needs to talk about this.
- Competition brings down prices, and chatbots used for support can be considered our competition.
- The clinicians most certain AI can’t replace us may be the ones who’ve never really tried it.
I took a continuing education class yesterday in which a colleague assured the audience that AI won’t take our jobs. It wasn’t the first time I’ve heard this in a CE setting. And each time it happens, I find myself more stunned than the last.
Source: Image created by ChatGPT
AI’s Impact on Earnings and Career Sustainability
Let’s start with the most common one: You’ll keep your job. Maybe. But keeping your job and having your job remain untouched are two entirely different things. As AI therapy advances, it creates genuine market competition. In every market competition drives prices down.
Thus, at minimum, the inevitable impact of AI on our profession will be reduced fees, but also likely the respect our degrees garnish when we are competing with an algorithm. Many young clinicians graduate with substantial debt, and this salary reduction changes the economics of entering this field entirely. The argument that therapists won’t lose jobs sidesteps a harder question: Whether the jobs that remain will be sustainable or well-compensated. If salaries decline, the cost of training for this profession has to change accordingly. That conversation is not happening.
Will Clients Prefer Human or AI Therapists in the Future?
The second reassurance is equally fragile. The argument is, there are so many untreated people who need mental health care, and so few therapists, that AI will simply expand access rather than replace us. There’s some truth here. But this argument has a hidden assumption that deserves examination.
For humans to maintain a meaningful role in the profession, three things have to be true simultaneously. A sufficient number of clients must prefer a human therapist over an AI, must prefer it enough to pay dramatically more for the privilege, and must be willing to tolerate the scheduling challenges that inevitably arise when scheduling with a human. A chatbot provides unlimited and completely flexible therapeutic interaction for roughly twenty dollars a month. A human therapist costs anywhere from one hundred fifty to three hundred dollars an hour and has scheduling limits such as limited evening and weekend time. Even people who genuinely prefer human connection have to prefer it at an extraordinary price and conveniance differential. That bar will get higher as AI improves.
And consider this: a 2025 APA peer-reviewed study by Rousmaniere and colleagues surveyed nearly 500 Americans with ongoing mental health conditions who had used AI in the past year. When comparing LLM’s to human therapy, almost 38% found LLM’s more helpful than traditional therapy. While only one study, it nonetheless deserves attention.
Key Findings Comparing AI and Human Therapy
The research is moving faster than the profession wants to acknowledge. Just months ago, a randomized controlled trial published in JAMA Network Open compared AI therapy, face-to-face group therapy, and a waitlist control in nearly a thousand college students. AI outperformed group therapy on anxiety and beat both groups on well-being and life satisfaction.
Again, one study proves nothing. But here’s what the profession keeps missing: this is a moment in time. AI is improving at a pace humans can’t. The question is not whether today’s chatbots match the best human therapists. The question is where the trajectory leads. And the trajectory is not ambiguous to me. Like the vast majority of occupations, chatbots will transform our profession and challenge our livelihoods.
How Important Is Perceived Therapeutic Alliance?
Then there is the therapeutic alliance argument. The relationship is what heals, the thinking goes, and AI cannot form a real relationship. Yet people anthropomorphize AI and report feeling genuinely connected to chatbots. The JAMA study found that perceived therapeutic alliance with the AI platform predicted engagement, which predicted symptom improvement. Furthermore, a substantial number of people actively prefer AI for mental health support precisely because it isn’t human. For clients feeling shame or who are fearful of judgement, a kind and gentle algorhythm can feel more soothing than a person.
There is also a disruption that we aren’t tracking at all: What chatbots do best right now is help the “worried well.” This is the same population newly minted-therapists learn on. Assuming many opt for nearly free AI care, our youngest professionals will lose their training ground. Clinical expertise is built incrementally. A profession where the lower end of the acuity spectrum disappears will struggle to develop the next generation of competent specialists. Again, nobody is talking about this.
Cost-Effectiveness May Be the Bottom Line
Insurers are watching all of this. Whether human therapists remain economically viable in a landscape where AI demonstrates comparable outcomes for common presentations is not a question payers will answer sentimentally. They will follow cost-effectiveness data. And we may already be seeing the first signs. Recently, one major insurer quietly reduced reimbursement rates for doctoral-level therapists to the level of master’s-prepared clinicians, effectively erasing the financial return on years of additional training. Most therapists I’ve spoken with had no idea this happened. That’s a problem. By the time these shifts become impossible to ignore, the window for responding will have narrowed considerably. If you still question this logic, I’m guessing you haven’t actually used a chatbot to work through something that genuinely matters to you. Whenever I do so, I’m shocked by how heard and seen I can feel.
Nonetheless, the problems with AI therapy are real and serious. Chatbots are sycophantic by design, telling users what feels good rather than what is true. They make things up, and they do it confidently, without any signal that something has gone wrong. They work outside the ethical and legal accountability structures that govern licensed practice. And whatever privacy they promise is largely illusory.
We Need to Grapple with AI Disruption Now
I am not at all enthusiastic about the upcoming changes to our profession, nor am I suggesting we capitulate and hand our clients over to machines. I am arguing that our denial is not a winning strategy, and we can’t solve a problem that we don’t understand. The therapists who will navigate this well are the ones willing to look at it clearly now, before the transformation occurs.
We encourage our clients to tolerate uncertainty, adapt to change, and face hard truths rather than avoid them. Let’s take our own advice. We know better than anyone that denial is not a coping strategy. The therapists who will thrive in the next decade are not the ones who were reassured in a CE class. They are the ones who see the change coming, ask hard questions, and adapt. That process starts with honesty with each other, and with ourselves. I’m not a professional seer, and it’s possible I could be wrong. But we can all agree that somethings afoot, and I suggest we pay attention.
References
Rousmaniere, T., Zhang, Y., Li, X., & Shah, S. (2025). Large language models as mental health resources: Patterns of use in the United States. Practice Innovations. Advance online publication. https://doi.org/10.1037/pri0000292
Shoshani A, Gurfinkel B, Kor A, et al. (2026). Efficacy of a Conversational AI Agent for Psychiatric Symptoms and Digital Therapeutic Alliance: A Randomized Clinical Trial. JAMA Network Open; 9(4):e266713. doi:10.1001/jamanetworkopen.2026.6713
