ChatGPT 의대수업 활용법 (2024년 3월)

English summary

This video describes practical strategies for using large language models and conversational AI tools such as ChatGPT, Claude, Gemini, and Copilot in university and medical education, including course administration, curriculum design, assessment, and educational reporting. It presents prompts as combinations of instruction, context, and guidelines, and recommends iterative prompting in English followed by translation into professional and technical Korean.

Key points

• Syllabi and lecture outlines can be generated by specifying the learners, course duration, required topics, time allocation, prerequisite knowledge, and reference documents.
• AI tools can support explanations of complex concepts, including comparisons of hepatic clearance models, stepwise descriptions of proton pump inhibitor pharmacokinetics, and analogies paired with clinical examples.
• The video describes generating anticipated student questions, engaging facts, rapid true or false questions, multiple choice questions, and rationales. Instructor review and modification of AI drafted assessment items remain necessary.

Audience

The intended audience includes university professors, medical educators, instructional designers, educational administrators, researchers, and advanced learners interested in AI integration in medical education.

Scope and limitations

The source argues that conventional essays asking students simply to describe or summarize a topic may have limited educational value because large language models can produce them readily. It proposes critical evaluation of AI outputs, inquiry driven investigation, and flipped classroom activities combining AI assisted preparation with face to face discussion. Other applications include provisional grading of short answer responses using structured rubrics, automatic classification of examination questions, and synthesis of qualitative course evaluations. The stated risks include hallucinations, run to run output variability, and false accusations arising from errors in AI detection software. Direct prompting in Korean is described as potentially slower and less precise than prompting in English. The supplied material is an evidence note rather than a reported empirical comparative study and provides no quantitative educational outcomes. Tool features and pricing may also change over time.

Research context

The topic has a neutral thematic connection to Sangzin Ahn's publicly presented interests in medical education, generative AI, and data informed educational practice.