Summarized by Raghav Wusirika, MD
Associate Editor, CE News
Virtual Research Roundtable – AI: Friend or Foe in CPD Research
Webinar presented on July 15, 2025
Presented by: Eleftherios (Terry) Soleas, Ph.D
Director of Continuing Professional Development at Queen’s Health Sciences
Artificial Intelligence (AI) has been a hot topic at my institution as well as globally, related to various aspects of our professional and personal lives. The goals of the session were to: raise awareness of the ethical principles guiding AI use in education; share practical examples of how AI can be applied to improve your work. This presentation aimed to define the best use cases for AI in CPD research. As Terry states at the beginning of the webinar, AI is both Friend and Foe depending on the situation. I found the opening statement to provide a great grounding principle for AI – “It replaces tasks, not people – and certainly not their humanity.” AI must be used “ethically, effectively, and efficiently” and must be used in a controlled manner to avoid biases and weaknesses.
Terry highlighted various use cases, including documentation, curriculum development, assessment, and creative tasks. Specific examples included image creation, learning objectives based on topics, grammar checks, creation of multiple-choice questions, and generation of educational scenarios. Still, these are done with varying levels of success. Of note, the educational scenarios created for case studies, when reviewed by actual physicians, were found to be riddled with errors. AI images are helpful but need to be repeatedly refined, as first drafts can have biases or contain extraneous details.
Comparatively Deep Research used to create overviews or tables for a literature review can decrease the time for that application. Verification of the references is necessary, as hallucinations are a risk, so the efficiency increase is not going to be as dramatic as initially hoped. Similarly, it can be used for statistical calculation, but again, the formula for this needs to be checked and understood so that the AI is not used as a black box that cannot be checked. It appears that the most straightforward use is in manuscript editing*, formatting, and adjusting reading level.
Ethics now require the use of AI to be disclosed in journal submissions. AI does run the risk of plagiarism in addition to its risk of hallucination. Although AI can use all sources available to it for background research, the use of targeted papers increases the yield. This also decreases the risk plagiarism when AI “summarizes” papers which it can only access an abstract for. Those summaries frequently appear the same as the abstract given the limited text available.
An advanced use case was featured in this webinar involving the creation of bots. In particular, it was a case study where learners can interact with a chatbot. The chatbot did allow learners to decide how to manage medical cases in an interactive way and were available on the participants own time which was valuable. Chatbots start as fairly rudimentary language models but will improve with repeated interaction. As a final aside, this short talk was worth a watch for the content but also the delivery. Terry made a potentially dry and nerve-wracking topic very entertaining. In closing, he summarized his view of how AI offers remarkable promise to augment and accelerate education provided it is used thoughtfully, centers on human achievement, and follows strict guidelines for privacy, confidentiality, and integrity.
*AI was used to edit this document about AI
