Study reveals an “Intention–behavior gap” in clinical teachers’ adoption of AI-assisted teaching

Artificial intelligence is quickly moving into medical education. Hospitals and medical schools are exploring how AI can help generate cases, prepare teaching materials, design assessments, and support more personalized learning. But for clinical teachers, being open to AI does not necessarily mean using it in day-to-day teaching.

A new study published in the Chinese Journal of Medical Education Research highlights that gap. Researchers from Peking University First Hospital conducted a questionnaire survey among clinical teachers at a university-affiliated hospital, distributing 359 questionnaires and collecting 335 valid responses and found that teachers reported relatively high willingness to adopt AI-assisted teaching, but much lower levels of actual use. The finding points to a practical challenge for medical education: AI tools may be widely discussed and increasingly accessible, but their integration into clinical teaching still depends on trust, support, and usable implementation pathways.

The study was conducted from February to April 2025. Using an extended version of the Unified Theory of Acceptance and Use of Technology 2, or UTAUT2, the researchers examined factors that may shape clinical teachers’ intention to use AI-assisted teaching and their actual usage behavior. In addition to established technology-adoption variables such as performance expectancy, effort expectancy, social influence, facilitating conditions, price value, hedonic motivation, and habit, the study also added perceived risk and cognitive trust to reflect the particular concerns surrounding AI in medical education.

The results show that AI is already part of many teachers’ professional and personal lives, but not yet a routine part of teaching. Among the 335 respondents, 78.81% reported using AI in daily life, 68.96% in research, and 60.30% in clinical work. Only 24.48%, however, reported using AI specifically for teaching. This contrast raises a central question: why does AI appear to enter life, research, and clinical practice more readily than clinical teaching?

The answer may lie in the difference between willingness and behavior. Clinical teachers’ mean adoption intention score was 3.93 on a five-point scale, while their mean usage behavior score was only 2.30. The difference was statistically significant, indicating a clear intention–behavior gap. In other words, many clinical teachers are willing to consider AI-assisted teaching, but that willingness has not yet translated into frequent or sustained classroom use.

Structural equation modeling showed that several factors significantly influenced adoption intention. Cognitive trust had the strongest effect, with a standardized path coefficient of 0.573. This suggests that teachers’ willingness to use AI depends heavily on whether they believe the tools are reliable, professional, and understandable. The result is important because it shifts attention away from technology promotion alone. For clinical teachers, the key issue may not be whether AI is being advertised as powerful, but whether it can be trusted in a high-stakes educational environment.

The study showed that cognitive trust was the strongest predictor of adoption intention. At the same time, hedonic motivation, effort expectancy, habit, performance expectancy, and facilitating conditions also had significant positive effects. Notably, although facilitating conditions had a relatively small standardized coefficient, its statistical significance was high, indicating that training, resources, technical support, and institutional environment remain important foundations for implementing AI-assisted teaching. This means that teachers care not only about whether AI is useful and trustworthy, but also about whether it is easy to use and whether their institutions provide sufficient support.

Some expected factors did not show significant direct effects. Social influence, price value, and perceived risk were not significant direct predictors of adoption intention in the model. This does not mean that teachers are unaffected by peers, unconcerned about cost, or indifferent to risk. Rather, within this study’s sample and model, those factors did not directly explain teachers’ willingness to adopt AI-assisted teaching. Trust in the tools themselves was far more influential.

The study also identified two independent indirect pathways between performance expectancy and adoption intention: one mediated by hedonic motivation and the other by cognitive trust. In other words, when clinical teachers believe that AI can help improve teaching efficiency and effectiveness, this perception may not only directly strengthen their intention to adopt AI-assisted teaching, but also further promote adoption by improving the user experience and enhancing trust in the tool. Put differently, whether AI-assisted teaching can be accepted by teachers is not only a question of whether it is useful, but also whether it is easy to use, trustworthy, and suitable for sustained use.

For hospitals and medical schools, the findings carry a clear message: encouraging AI-assisted teaching requires more than enthusiasm, policy slogans, or occasional tool recommendations. Institutions need credible systems for evaluating AI tools before they are introduced into teaching. Such systems should assess educational effectiveness, data security, transparency, and the boundaries of appropriate use. Making those evaluations visible to teachers may help build the trust needed for adoption.

Training also matters. Teachers may need concrete examples of how AI can support clinical teaching, rather than broad claims about its potential. Early use cases could focus on lower-barrier tasks, such as generating quiz questions, summarizing literature, preparing teaching materials, or drafting preliminary learning activities. These practical entry points may help teachers gain experience with AI while still maintaining professional judgment over content and pedagogy.

The authors also point to the importance of institutional support. If clinical teachers are expected to use AI-assisted teaching, they need access to reliable tools, technical assistance, training opportunities, and enough flexibility to experiment. Without these conditions, teachers may remain interested in AI but unable to convert that interest into regular practice.

The study has limitations. It was conducted at a single teaching hospital using convenience sampling, which may limit the generalizability of the findings. The sample also had certain demographic concentrations, including a higher proportion of women, teachers aged 30 to 39, and internal medicine teachers. Because the study was cross-sectional, it captured relationships at one point in time and cannot establish how adoption intention and usage behavior change over time. Future multicenter and longitudinal studies will be needed to examine whether these patterns hold in other clinical teaching settings.

Still, the study offers a timely look at the realities of bringing AI into medical education. Clinical teachers may be willing to use AI, but willingness alone will not carry the technology into teaching practice. Trustworthy tools, practical guidance, institutional support, and visible teaching value appear to matter most. For AI-assisted teaching to move from promise to routine use, medical education leaders may need to focus less on persuading teachers that AI is the future, and more on building the conditions that make responsible use possible.

English Title: Analysis of adoption intention and usage behavior of clinical teachers regarding artificial intelligence-assisted teaching using the UTAUT2 model

Journal: Chinese Journal of Medical Education Research

DOI: 10.3760/cma.j.cn116021-20251022-02219

Authors: Qiao Jing, Han Wei, Li Ke, Qi Jianguang

Institution: Department of Education, Peking University First Hospital

Chinese Journal of Medical Education Research is a monthly peer-reviewed journal sponsored by the Chinese Medical Association and hosted by Chongqing Medical University, under the supervision of the China Association for Science and Technology.
Launched in 2002, the journal publishes research and practice-oriented studies on medical education, with a focus on teaching reform, clinical teaching, curriculum development, residency training, graduate education, nursing education, educational technology, and international medical education. It also features themed columns and special issues on emerging topics and institutional innovations in medical education.
The journal is recognized as a Source Journal for Chinese Scientific and Technical Papers and Citations and is included in the Chinese Science and Technology Core Journals. It is indexed in Wanfang Data, Index Copernicus, the WHO Western Pacific Region Index Medicus, and Ulrich’s Periodicals Directory.

Journal page: http://yxjyts.alljournals.ac.cn/homeNav?lang=zh