AI Resilience Report forTherapists, All Other

Right now, AI is mostly augmenting therapists rather than replacing them. "I have not seen within mental health care any jobs be replaced by AI as of yet," says Vaile Wright of the American Psychological Association. Instead, the growing adoption of AI in mental health care has been mostly limited to certain kinds of tasks — primarily paperwork like session notes, billing, and electronic health records ^[1]. Researchers at the University of Utah recently described four categories along a continuum: scripted chatbots, AI that evaluates therapists, AI that assists therapists with suggestions while a human delivers care, and autonomous AI agents that provide therapy directly, noting that most safe uses today are the lighter ones like note-taking ^[2].

At the same time, clients are doing their own augmentation: a KFF poll found 16% of adults — and 28% of those 18–29 — used AI for mental health advice in the past year ^[3], and clinicians have started asking every patient about their AI use to better understand what's "inside their patient's brain", according to reporting by WBUR ^[4].

Adoption is moving faster than many therapists expected. There are nearly 40 different products offering transcription and documentation support for providers, and large systems are piloting intake bots — Kaiser Permanente is evaluating the chatbot Limbic, prompting a one-day strike by 2,400 clinicians worried about job erosion ^[1]. Cost pressure, a national therapist shortage, and 24/7 demand push systems toward AI; meanwhile, uninsured adults are more than twice as likely as insured adults (30% vs. 14%) to use AI for mental health ^[3], showing real unmet need.

But brakes exist. LLMs carry huge risks since they are known to fabricate information, encode biases, and respond unpredictably, and the ACA is rewriting its Code of Ethics to directly address AI, telehealth, and emerging technology ^[5], signaling that licensure and liability rules will shape how fast AI enters the therapy room. The good news for you: empathy, trust, crisis judgment, and ethical decision-making — the deeply human parts of this work — remain the parts no chatbot has convincingly replicated.