Presentation
22. Teaching ChatGPT: Attempting to Demonstrate the Protégé Effect With a Large Language Model Learner
SessionPoster Session 2
DescriptionIt is well known that teaching others can benefit the teacher's understanding. This is commonly known as the "learning-by-teaching" or "Protégé" effect. The effect is leveraged in systems such as the "see one, do one, teach one," approach, common to medical training. However, its applicability is limited when no additional learner is available. Addressing this limitation, researchers have explored non-interactive learning-by-teaching methods (such as simply imagining the learner) with mixed outcomes. As technology advanced, focus shifted to using teachable software agents to substitute for human learners. Teaching the agents was shown to be beneficial, but most were restricted to predefined content. Recent advances in artificial intelligence have enabled more open-ended interactions with conversational agents. This study was an investigation of whether a general-purpose chatbot, ChatGPT, can serve as a teachable agent and replicate the learning-by-teaching/Protégé effect. The effect was not replicated but the study design is informative for future efforts.
Event Type
Poster
TimeThursday, September 12th5:30pm - 6:30pm MST
LocationMcArthur Ballroom
Aging
Augmented Cognition
Children's Issues
Communications
Cybersecurity
Education
Environmental Design
General Sessions
Human AI Robot Teaming (AI)
Macroergonomics
Occupational Ergonomics
Student Forum
Surface Transportation
Sustainability
System Development