When AI enters our classrooms: the dynamics of the “AI and Health Education” working group
A collaborative dynamic to transform our teaching practices
Launched on March 6 during the Commission for Studies and Student Life (CEVU) meeting, the “AI and Health Education” working group is tasked with coordinating training and the use of artificial intelligence tools in education at the Sorbonne University Faculty of Health Sciences. In six months, this working group has mobilized the university community and initiated a reasoned integration of AI into our teaching practices.
An assessment revealing needs and expectations
To establish a shared diagnosis, all heads of teaching departments were asked to join the working group and to complete a questionnaire aimed at mapping current and desired uses of AI tools, particularly large language models (LLMs), for the creation of educational content.
This questionnaire received 67 responses representing 27 specialties (radiology, surgery, pulmonology, pediatrics, public health, general medicine, emergency medicine, medical genetics, dermatology, etc.). In the results, 45% of respondents reported having already used AI tools in education (ChatGPT Copilot, DeepSeek, Claude, Consensus, and image generators). Current uses mainly involve illustrating courses, automated assessment, improving slideshows, proposing lesson plans, and even writing code for research. Regarding future prospects, respondents expressed a keen interest in diversifying their uses: writing course materials, creating exam questions and grading rubrics, developing interactive clinical cases, self-assessment tools, and lesson preparation were among the most popular applications. The expectation for support is particularly strong: among all participants, the desire to benefit from
Concrete, collaborative projects
At the first meeting last June, and then at the September session, several AI projects developed within the faculty were presented, demonstrating the vitality of educational innovation at our institution.
The EVAL-IA project, led by Dr. Franck Verdonk, proposes a system for automatically recording and analyzing student transcripts during ECOS exams, with the aim of improving the objectivity and reliability of grading. Dr. Quentin Vanderbecq and Dr. Pierre-Antoine Soret presented their work as part of the teaching diploma, using an LLM for the automatic grading of short open-ended questions (QROC), with validation by three human annotators. Finally, Dr. Fréderick Gay presented the results of an automatic analysis of student responses to a satisfaction questionnaire conducted at the faculty, carried out via the creation of a local search engine that allows these data to be exploited efficiently.
Training and equipping the community
At the same time, a team of educational engineers from the SPN, SCAI, and BU has developed two training workshops, scheduled for November and December in person, which will be offered again in 2026. The first workshop will focus on raising awareness of the challenges and possibilities of AI in education. The second workshop will provide practical experience and hands-on training in the use of the tools.
In addition to these in-person workshops, a series of video tutorials is being produced for academics, based on very concrete use cases: how to write course questions that comply with the principles of docimology using course materials with an online agent? How can you automatically create an ECOS station? How can you develop a search engine for your own courses or recommendations? These resources will allow everyone to progress at their own pace in learning how to use these new tools.
Outlook
Six months after its launch, the “AI and Health Education” working group is continuing its momentum with the aim of promoting the thoughtful, critical, and creative integration of AI into our teaching practices. The support of colleagues and the multitude of projects underway confirm the interest in these tools, provided they are used in a thoughtful and guided manner.
The coming months will see continued support for teachers to develop their skills and the dissemination of best practices within the Faculty. Discussions are also underway regarding the adoption of a common, secure LLM adapted to educational needs, in a responsible approach combining educational effectiveness and digital sobriety.
For further information, please do not hesitate to contact us: christel.gerardin@aphp.fr
Dr Christel Gérardin
Coordination of the “AI and Health Education” Working Group – Faculty of Health Sciences, Sorbonne University