The tools of the future are the challenges of today: The use of ChatGPT in problem-based learning medical education

Med Teach. 2024 Mar;46(3):320-322. doi: 10.1080/0142159X.2023.2290997. Epub 2023 Dec 27.

Abstract

What is the educational challenge?Incorporation of large language model (LLM) or generative artificial intelligence (AI) software poses a challenge to various areas of medical education, including problem-based learning (PBL). LLMs, such as ChatGPT, have incredible potential to transform educational systems and enhance student learning outcomes when used responsibly.What are the proposed solutions?ChatGPT can provide several ways to support students and assist facilitators with course responsibilities. Here we address factors of implementation and describe how ChatGPT can be responsibly utilized to support key elements of PBL.How was the solution implemented?Providing reasonable access is an essential element of novel software implementation. Additionally, training for both faculty and staff is vital to foster responsible usage, provide base-line proficiency, and guide users to critically evaluate the quality of output.What lessons were learned that are relevant to a wider audience?The use of LLMs or other generative AI is dramatically rising in the world. Appropriate and conscientious incorporation of AI into educational programs can foster responsible use and potentially enhance student learning.What are the next steps?Assessment of learning outcomes, student self-efficacy, group dynamics, and stakeholder feedback are required to measure the effects of ChatGPT in the PBL curriculum. Additionally, software programs competitive with ChatGPT are currently under development and will also need to be investigated for their potential role in education.

Keywords: ChatGPT; generative artificial intelligence (AI); large language models (LLM); medical education; problem-based learning (PBL).

MeSH terms

  • Artificial Intelligence
  • Curriculum
  • Education, Medical*
  • Humans
  • Learning
  • Problem-Based Learning*