Everyone already knows that artificial intelligence is currently a hot controversial topic. In the medical community, it has as many proponents as enemies. Especially the latter have been wondering if AI will soon replace physicians and teachers of medicine. This question was also posed during the symposium that took place at the MUW as part of the University Quality Culture Day.
The meeting attracted academic teachers, students, AI experts, and guests from Ukraine. It was hosted by the representatives of the MUW University Taskforce for the Quality of Education: Piotr Dziechciarz, PhD Hab. (chairman), and Antonina Doroszewska, PhD, as well as Jakub Sokolnicki, PhD, Head of the Center for Education Quality and Competence Development at the MUW. The conference took place at the Medical Simulation Center, but it was also streamed online.
Artificial intelligence in research
Is it acceptable to use ChatGPT when writing and reviewing academic papers? Where is the line? With these questions, Professor Hanna Szajewska, Director of the MUW Department of Pediatrics, started her presentation titled “AI in academia: artificial intelligence, real challenges”.
First, she presented the principles that govern the use of AI when writing research papers.
“Those rules differ from journal to journal. Many editorial boards require that authors disclose the use of AI for their papers,” she said.
The same is true when reviewing papers. For example, “Nature” requires that the use of AI be described in detail. Authors have to provide details of the system they used, the prompt, the date when they asked the question – all this must be publicly available. But another journal, “Jama”, does not accept the use of AI for writing reviews at all.
Professor Szajewska admits to using ChatGPT when reviewing her own papers.
“Of course, with some of its comments I do agree, while with others I don’t. But it would be unwise of me to disregard this completely. AI is often perceived as a form of fraud, whereas it could be seen as a tool that supports science and research,” emphasized the expert.
The Professor also discussed other AI tools and their use in medicine, mainly in imaging. And she concluded her presentation by quoting Jesse M. Ehrenfeld, President of the American Medical Association: “AI will never replace physicians. But physicians who use AI will replace those who don’t.”
Artificial intelligence in teaching
Joanna Mytnik, PhD Hab., Director of the Center for Innovative Education at the Gdańsk University of Technology, discussed the use of AI in teaching. She presented a talk titled “Global change in thinking about learning: AI and Generation Alpha at universities”. The expert characterized Generation Alpha, i.e. the current teenagers, who will reach the university as early as in 2030. She emphasized that those people learn through constant interaction, and in doing so they focus more on the results than on the process. Her belief is that this poses a significant challenge for teachers.
Modern teaching methods that use artificial intelligence and their practical applications were discussed by Agata Komendant-Brodowska, PhD, from the Department of Digital Sociology at the University of Warsaw. The title of her speech: “We are the ones carrying out digital transformation – how to include new methods in your teaching practice and stay sane”.
The University Quality Culture Day also included a presentation by Marcin Kaczor, PhD, Director of the MUW Medical Simulation Center and an enthusiast of new technologies. He started by highlighting the dynamic growth of the field of medical simulation. In 2014, Poland had four medical simulation centers, while in 2024 there were already 92 of them.
He also discussed the benefits of using AI in medical education. Those include the personalization of learning, the possibility to create virtual assistants and chatbots that could function like teachers, as well as automatic evaluation and feedback, generating virtual patients and clinical cases, and improving engagement through gamification (the use of games in teaching).
“We have opportunities, but we also need to bear the limitations in mind,” he observed. “Those include the risk of errors, risk of data bias, and the lack of full reliability.”
The use of AI in medicine and teaching was also the topic of a presentation by Anna Rosada, MD, from the Medical Communication Department. Anna Kupczak, president of the MUW Student Government, talked about her experience with ChatGPT, and demonstrated the results of her survey among 50 students of our university, which concerned the use of artificial intelligence, mostly large language models.
A practical tool for generating questions
Professor Joanna Gotlib-Małkowska, Head of the MUW Department of Education and Research in Health Sciences, presented an especially interesting speech titled “MedQuiz Professional Designer: quick, good, and effective question design”. She noted that an increasing number of papers on the use of artificial intelligence in education have been appearing on Pub-Med. There include studies comparing various language models, and you can also find papers demonstrating how ChatGPT fares with different writing systems (it fares best in English).
“There is also a paper by Polish authors who analyze how well ChatGPT deals with the medical final examination in Poland (LEK). For a year now, there has also been interest in analyzing exam questions generated by ChatGPT,” said the expert.
Professor Gotlib-Małkowska presented an interesting study that compared the level of difficulty and differentiating power of questions designed by a human and those created by artificial intelligence. It turns out that when it comes to the level of difficulty, there are indeed statistically significant differences. But when we take into account the differentiating power (i.e. whether a question helps distinguish students who have mastered the topic well from those who struggle with it), there were no significant differences.
“But we need to remember that the quality of questions designed by AI depends to a great extent on the quality of the prompt and the additional context,” emphasized the expert.
Professor Gotlib-Małkowska is the author of MedQuiz Professional Designer. It is a model built on the basis of ChatGPT and available within that application. It is intended to support academic teachers, but it can also be used by students. It devises test questions. The expert demonstrated how it works step by step.
“First, we need to specify how many questions we want to generate, and of what type: should they be single-choice questions, gap-filling questions, or more complex ones? The model must also be told whether it should generate the answers. Finally, we add context, that is a file with the teaching material based on which Chat should create the questions. Once we have input that information, we will get the questions. Obviously, you always need to verify their accuracy,” she emphasized.
Professor Gotlib-Małkowska’s presentation sparked many questions, including about the copyright of the context fed into ChatGPT.
“Every good symposium provides good answers to many questions, but generates even more of the latter. And in that respect, our meeting has fulfilled its role perfectly well,” concluded Piotr Dziechciarz, PhD Hab.
The event was also attended by Professor Rafał Krenke, MUW Rector, and Professor Aneta Nitsch-Osuch, Vice Dean for Curriculum and Educational Quality at the Faculty of Medicine.
