Park and Kim's November 3 student survey, WHO's existing AI guidance, and AOTA's Education Summit program
Occupational therapy students surveyed at three South Korean universities want AI teaching connected to clinical reasoning. A study published November 3 in the Journal of Multidisciplinary Healthcare puts that connection at the top of their educational-content ratings.
In brief
- The survey describes preferences and perceived ability, without testing clinical performance.
- A classroom exercise could examine an AI recommendation against a fictional client's circumstances.
- Educators would still need to evaluate whether the lesson improves students' decisions.
What the students reported
Yun-Hee Park and Deok-Ju Kim analyzed 230 complete responses collected in July and August. Students rated their AI competence at 3.42 out of five and educational-content needs at 3.93. Hands-on digital rehabilitation work was their favored teaching method; client-appropriate tool selection drew the lowest competence rating.
The convenience sample and researcher-developed questionnaire limit generalization. The instrument lacks full psychometric validation, and a one-time survey cannot establish educational effectiveness. The authors also report no prospective IRB approval or formal exemption determination. The findings offer a starting point for local questions about teaching.
Give the recommendation a client
For an educator, one possible response is to build a lesson around a fictional client trying to prepare breakfast after a stroke. Give students an AI-generated activity plan and ask them to examine its assumptions. Does it account for the person's kitchen, fatigue and goals? Which claims have evidence behind them? What would require further assessment before a recommendation could be used?
That exercise would need its own evaluation. Faculty could compare students' reasoning before and after instruction, looking for whether they detect unsupported claims and explain their revisions. A polished activity plan gives an educator little evidence of learning unless the student can defend the decisions inside it.
Keep oversight in the lesson
WHO's existing 2024 guidance describes risks from inaccurate or biased generative AI output and from people overlooking errors when they rely on a system. It also identifies threats to patient information. Those concerns give faculty concrete subjects for feedback: the quality of a cited source, an unsupported inference, and the information a student proposes to enter into a tool.
The discussion is already on the U.S. education calendar. AOTA lists AI in curricula and clinical training among the subjects for its November 13–14 Education Summit in Indianapolis. Programs reviewing their own teaching can begin with a modest question: can a student explain why a technology fits this client's occupation, and recognize when the recommendation needs to stop?