Clinician Competence in the Age of Chatbots

Patients are adopting new technologies faster than clinicians are prepared to respond. Millions of people already use large language model chatbots such as ChatGPT for emotional support, psychoeducation, and informal therapeutic guidance.¹ The mental health community recognizes that technology is reshaping care. Clinicians have previously adjusted to telehealth, electronic health records, and internet-mediated relationships. Chatbots introduce something different because they sit closer to the core of clinical work, shaping how people reflect, regulate, and make meaning of their experiences, often before they ever enter the therapeutic space.

Clinical competence and training must evolve. Today's clinicians need literacy in chatbot-mediated support to understand how patients engage with these tools. Without this, clinicians risk losing their role as effective guides in rapidly changing psychological environments. At the same time, AI chatbots may enhance aspects of the therapeutic process, which means clinicians must also understand how they can be used safely and appropriately within care.

Here we discuss core competencies that training programs and practicing clinicians should begin developing now. These are ever-changing, but creating dialogue and frameworks is imperative to training competent clinicians.

Conceptual Literacy About Chatbots

Clinicians should understand how generative AI systems work and the risk they pose to privacy. This includes recognizing that chatbots generate probabilistic language outputs rather than possessing insight, memory, or genuine relational understanding. Such conceptual clarity allows clinicians to help patients interpret chatbot interactions realistically and avoid both over-trust and unnecessary fear. Clinicians should also have insight into which tools might best fit patients' needs and privacy requirements. In practice, clinicians can build this literacy through continuing education courses, reading current research on LLMs in mental health, and experimenting with chatbot tools directly. Training programs can embed AI-focused modules into foundational coursework and assign hands-on exercises that compare chatbot outputs with evidence-based interventions.²

Assessment of Chatbot Use in Clinical Intake

Just as clinicians routinely assess sleep, substance use, or social media behavior, asking about chatbot use is becoming increasingly relevant. Understanding how clients use chatbots can provide valuable clinical context. In practice, clinicians can add structured questions about AI tool use to existing intake forms and screening protocols. Training programs can incorporate chatbot assessment into clinical interviewing courses and role-play exercises, ensuring trainees learn to explore this domain with the same rigor as other behavioral assessments.

Risk Management and Safety Awareness

While chatbots can provide comfort, they can also deliver inconsistent or misleading guidance. Clinicians should be able to discuss limitations openly and help patients develop safety strategies when using these tools, including clarifying when AI support may be helpful and when human intervention is essential. In practice, clinicians can stay current on documented chatbot failures and safety incidents and develop informed consent language that addresses AI use. Training programs can integrate AI risk scenarios into ethics coursework and clinical supervision, teaching trainees to identify and respond to AI-related clinical risks in real time.

Integration Into Case Formulation

Chatbot interactions can shape cognition, emotion regulation strategies, and expectations about relationships. For some patients, AI may function as a coping tool, while for others, it may reinforce avoidance or dependency. Clinicians need preparation to incorporate technology use into case conceptualizations. In practice, clinicians can begin documenting patients' AI use patterns alongside other relevant behavioral data and discussing these patterns in peer consultation. Training programs can include AI-informed case formulation in practicum seminars, using real or simulated clinical vignettes.

Skillful Use of AI as an Adjunctive Tool

Beyond patient use, clinicians themselves may employ AI for tasks such as psychoeducational material generation, session planning, or reflective practice. Competence involves using these tools thoughtfully without outsourcing clinical judgment or ethical and privacy responsibility. In practice, clinicians can experiment with AI tools in low-stakes professional tasks while developing personal guidelines for appropriate use boundaries. Training programs can offer supervised AI tool labs where students practice using AI for clinical support tasks and receive feedback on ethical application.

Ethical AI Supervision and Training

Supervisors play a critical role in shaping how the next generation of clinicians engages with AI. Clinical supervisors need training in how to guide supervisees through AI-related ethical dilemmas, safety considerations, and integration decisions. In practice, supervisors can incorporate AI-related case discussions into regular supervision sessions and seek out supervisor-specific continuing education training on AI in clinical work. Training programs can require supervisor certification curricula that include AI safety and ethics modules alongside traditional supervision competencies.³

Ongoing Reflective Adaptation

This field changes rapidly. Clinicians must adopt a practice of continuous learning. Technological change in mental health is unlikely to stabilize, and training programs that emphasize flexibility and curiosity will be best positioned to prepare therapists for sustained professional relevance. In practice, clinicians can join professional communities focused on AI in mental health, subscribe to relevant research feeds, and set regular intervals for reviewing their AI-related practices. Training programs can build ongoing technology review into their curricula and encourage students to develop habits of professional self-assessment around emerging tools.

Training Responsibility

Much of this burden currently falls on individual clinicians to navigate alone. Supervisors, trainees, and early-career clinicians encounter patients who already use chatbots for support, and they are often left to make sense of it without consistent, empirical guidance. If this is a real part of some patients' lives, then it must be a real part of how we train clinicians. Training programs should help clinicians think through what these tools are doing in people's emotional lives and how they might be used safely to improve patient care.

Developing these competencies will require new educational models and of course much more research. Programs across the country must increase their flexibility in incorporating new trainings. Many programs have already started doing this: one of the first to integrate was Sentio University, a nonprofit Master of Arts in Marriage and Family Therapy (MFT) program based in Los Angeles. Founded in 2020, the MFT program emerged from recognition that many graduate therapy trainees were not achieving strong clinical outcomes. While the program was not initially centered on AI, supervisors began noticing that clients were using chatbots for emotional processing. In response, Sentio developed an AI research initiative to study how these tools are being used and what they mean for therapist training and clinical practice. The MFT program focuses on the competencies that AI cannot replace while equipping students with education in how and when AI tools might safely aid treatment. They also offer supervisor training that includes AI safety and ethics education, preparing clinical supervisors to guide trainees through AI-related challenges in practice.⁴ Additionally, the university has published a free AI course for mental health professionals, recognizing that education on this topic should be widely accessible.

As many programs slowly implement new trainings, efforts like these begin to outline what structured competency training in an AI-influenced clinical landscape could look like. Guided by research, programs should be mindful of being balanced in their integration of these technologies. They should avoid following hype cycles and prioritize developing future proof mental health professionals.

As mental health care evolves, preparing clinicians for this new frontier will be necessary to ensure therapeutic integrity, advance clinical outcomes, and promote patient wellbeing.

Dr Frances is professor and chair emeritus in the department of psychiatry at Duke University.

Ms Noorily writes and works at the boundary between AI and the humanities.

References

1. Rousmaniere T, Zhang Y, Li X, et al. Large language models as mental health resources: patterns of use in the United States. Practice Innov. 2025.

2. About Sentio University. Accessed April 13, 2026. https://sentio.org/aboutsu

3. Clinical supervisor training in California. Sentio University. Accessed April 13, 2026. https://sentio.org/supervisor-training

4. Free AI course for mental health professionals. Sentio University. Accessed April 13, 2026 https://sentio.org/free-ai-course-therapists