Small Bytes Blog

AI in Mental Health Counseling: Education and Professional Training

By Lindai Xie, Ph.D., Educational Psychology, Leadership, & Counseling

The development of technology and education is deeply intertwined, with advancements in one domain often prompting adaptation and transformation in the other. Changes in environmental conditions, evolving learning needs, institutional regulations, and other external factors can initiate shifts in either domain, which in turn create a chain reaction that influences the other. For example, the COVID-19 pandemic significantly altered the educational landscape. The external environment, particularly social distancing restrictions, necessitated a rapid transition to online teaching. This shift accelerated the development and adoption of online teaching tools and platforms. In turn, these technological advancements required educators to develop new competencies, including familiarity with digital tools and the ability to design engaging and interactive virtual learning experiences. Therefore, it is not surprising that the rapid development of generative AI is transforming teaching and learning, fostering creative uses of AI tools to enhance teaching quality, and improving learning efficiency.

I began my exploration of AI through my research in the higher education field. Since late 2023, discussions surrounding AI in higher education have increased substantially. Early conversations primarily focused on individuals’ attitudes toward AI use. Consistent with prior research, my study found that overall attitudes toward AI integration are generally positive. Users value AI’s ability to generate human-like responses and assist with tasks such as email drafting and summarizing lengthy documents, while also recognizing its limitations, including concerns about accuracy and bias. As the conversation has evolved, educators and researchers have increasingly focused on developing policies and institutional guidelines to support ethical and appropriate AI use among students. Faculty have also emphasized the need for clearer guidance on how to effectively introduce AI into the classroom, as well as the importance of faculty-specific training to support responsible and meaningful integration.

Regarding the mental health field, particularly in counseling, people think we are behind. In my understanding, the discussions surrounding AI are ongoing in our profession; however, its adoption remains more limited compared to other disciplines in higher education. A primary factor contributing to this hesitation is ethical concern. Research in computer science and related fields has highlighted that AI systems may use user data to improve model performance, which conflicts with counselors’ fundamental ethical principles regarding confidentiality and the protection of clients’ personal information and privacy. While there are specific circumstances in which counselors may be required to break confidentiality, such as court orders or situations involving risk of harm to self or others, using client information to improve AI models does not fall within these exceptions. In discussions with counselors and counselor educators, one participant expressed concern about confidentiality risks, stating, “I know that I should not send identifiable information to AI, but I am afraid that if I forget to remove a client’s personal information, the AI may use that information to respond to another counselor’s question or generate case materials for student learning. That would constitute a breach of confidentiality.”

A breach of confidentiality is fundamentally different from the misuse of AI in academic settings, such as violating class policies or needing to redo a task. For counselors, confidentiality is closely tied to professional identity, ethical responsibility, and licensure. Violations can have serious professional and legal consequences. This helps explain why the adoption of AI in counseling remains more limited compared to other fields. Another reason is related to the current capabilities of generative AI. While these tools are highly effective in generating text-based content, and recent advances have improved their verbal and conversational abilities, they still fall short of replicating the depth of human interaction required in counseling. Given the importance of nuanced communication, emotional attunement, and therapeutic presence, the application of AI in counseling sessions is still in an early stage and requires further development before it can be widely integrated into practice.

However, I believe AI integration for counseling session preparation (e.g., session plan) as well as counselor education can be beneficial and worth practicing and trying. More specifically, for teaching in mental health, AI can be a teaching assistant. Faculty in counselor education can use AI to reduce the time typically required for mock session practice and supervision. In traditional counselor training, faculty are expected to observe mock counseling sessions with students in training and provide individualized feedback to support their professional growth and skill development. Another common practice involves pairing students for mock sessions, which are recorded and later reviewed by faculty for feedback. Faculty can develop AI-supported prompts that enable students to conduct mock counseling sessions with verbal AI and request performance-based feedback from the AI. This helps faculty save time by reducing the need to repeatedly deliver general feedback across multiple students, allowing them to focus instead on providing more targeted and constructive guidance, informed by students’ reflections on practice and any unresolved questions left by the AI. At the same time, students’ learning experiences are enhanced, as they can engage in repeated practice with AI at their own pace and receive immediate, personalized feedback that supports both skill development and self-confidence.

Although some students have already begun using AI independently and without formal guidance, faculty are encouraged to intentionally integrate AI into their teaching and introduce it using the proposed framework (see Figure 1 below). Students’ understanding and familiarity with AI can be categorized into three stages. In Stage 1, students are aware of AI but possess only limited, surface-level knowledge. At this stage, faculty should focus on introducing foundational concepts to enhance students’ AI literacy, awareness, and ethical orientation. In Stage 2, students have begun using AI but often lack guidance and a clear understanding of appropriate and responsible use. Faculty can support these students by providing opportunities for observational learning, such as modeling how AI can be used effectively, and facilitating guided discussions that allow students to learn from both faculty and peers. In Stage 3, students have developed foundational knowledge and are familiar with common AI applications. At this stage, they are ready to critically and effectively integrate AI into their work. Hands-on learning experiences are essential, along with continued supervision to support responsible and skillful application. Based on these stages of AI competency and the corresponding training needs, I developed a three-session AI workshop.

Figure 1. Students' AI Competency Stages and Corresponding Learning Needs

Flowchart describing Students’ AI Competency Stages and Corresponding Learning Needs

The workshop aims to strengthen counselors-in-training's confidence in using AI within a professional context. More importantly, it introduces how generative AI can scaffold student learning and develop scalable practices for incorporating AI into pedagogical design. This workshop seeks to enhance both instructional delivery and learning outcomes. The integration of AI is designed to assist students with varying levels of academic preparation, enabling them to engage more deeply with course content and apply their learning in practical, reflective ways. The workshop has been piloted with a group of students, and initial feedback has been positive, highlighting the need for structured AI training in the mental health field. One of my key takeaways from this pilot study is that, for students in programs where AI-related content is not a required component of the curriculum, the introduction of foundational AI knowledge should be simplified. Rather than relying on complex terminology, instructors should focus on explaining core concepts through concrete examples to enhance accessibility and understanding. In addition, when addressing the limitations of AI, it may be more effective to move beyond simply naming issues such as bias or transparency deficits. Instead, incorporating live or virtual demonstrations that directly illustrate these limitations can help students more clearly understand the practical implications of AI use.

This framework provides a systematic approach to fostering AI competency by integrating staged learning, ethical considerations, and guided practice. Although developed within the context of mental health counseling, it has broader applicability across disciplines. By tailoring implementation to students’ developmental needs and course design, educators can adopt this framework to facilitate meaningful and responsible AI integration in diverse educational settings.

By Lindai Xie, Ph.D., Educational Psychology, Leadership, & Counseling

Learn more

AI Teaching Resources

AI Tools for Research & Writing | University Libraries

The AI Small Bytes is a blog written by members of the AI Resources & Guidelines Committee. This blog will be updated periodically with new resources and information, and we hope that you will check back often. If you would like to talk about teaching with artificial intelligence and your concerns or ideas, please feel free to contact Lisa Low, Director of the Division of AI Pedagogy and Policy.