Abstract Purpose Artificial intelligence (AI) is expected to reshape medicine not only in clinical care but also in medical education. A key aspect of clinical training is the assessment of medical knowledge and its routine application in clinical practice. Traditionally, factual medical knowledge has been acquired through text-based formats, which are now widely accessible in digital form. However, many learners also prefer auditory formats. Today, AI offers new opportunities to generate such audio-based learning materials at scale and tailored to individual needs. We evaluated whether an AI-generated podcast can support knowledge acquisition comparably to reading the source publication. Methods An AI-generated podcast was produced from a peer-reviewed surgical publication. Participants (residents of orthopedic and trauma surgery) were randomized to either read the article or listen to the podcast. After the intervention, objective knowledge was assessed with a multiple-choice test; subjective knowledge and learning preferences were evaluated using a rating scale. Results The study included 15 residents (mean age: 26.5 years, range: 26–35). Baseline self-rated knowledge was 3.5 on a 1–10 scale. Of the residents, 13 identified primarily as visual learners, two as kinesthetic, and none as auditive learners. Ten participants (66%) reported feeling confident to perform the surgical approach after the intervention compared to five before the intervention. Mean test accuracy was 73% (range: 33–93%): 73% (range: 33–93%) in the podcast group and 73% (range 40–93%) in the reading group. Self-reported knowledge gain was greater in the podcast group (6 vs. 5 on a 1–10 scale). Conclusion This small exploratory randomized study found that AI-generated podcasts may support short-term knowledge acquisition comparably to reading the source material. Learner preferences vary and should be considered when designing educational resources. Larger, adequately powered trials with standardized outcomes are warranted to define the role of AI-generated audio files in medical education. Level of evidence: Level II.
Hockmann et al. (Tue,) studied this question.