Student and AI responses to physics problems examined through the lenses of sensemaking and mechanistic reasoning
Project Overview
The document explores the transformative role of Generative AI, specifically ChatGPT, in STEM education, with an emphasis on physics problem-solving. It analyzes both student and AI-generated responses to physics questions, utilizing the frameworks of sensemaking and mechanistic reasoning. Findings indicate that AI provides structured solutions that integrate formal and everyday knowledge, showcasing its potential to enhance physics education. However, student responses exhibit iterative reasoning and representation practices, suggesting that while AI can be a valuable tool, it also has limitations in fostering critical thinking and may expose gaps in students' knowledge. Overall, the study underscores the promise of Generative AI in education while also calling attention to the need for careful integration to ensure that students develop essential reasoning skills.
Key Applications
ChatGPT for physics problem-solving
Context: Physics education for introductory students
Implementation: AI responses were generated using a Zero Shot approach to prompt ChatGPT with physics problems.
Outcomes: AI solutions exhibited structured reasoning and identified forces acting on riders in the Gravitron problem.
Challenges: AI responses sometimes provided incorrect conclusions and lacked the ability to notice knowledge gaps.
Implementation Barriers
Cognitive Limitations
AI lacks the ability to notice gaps in its knowledge system and cannot engage in metacognitive reflection.
Proposed Solutions: Encourage students to critique AI-generated responses and enhance their understanding of physics concepts.
Representation Challenges
AI struggles to generate sophisticated diagrams and representations which are crucial for physics problem solving.
Proposed Solutions: Design assessments that encourage students to create diagrams and use visual aids to support their reasoning.
Project Team
Amogh Sirnoorkar
Researcher
Dean Zollman
Researcher
James T. Laverty
Researcher
Alejandra J. Magana
Researcher
Sanjay Rebello
Researcher
Lynn A. Bryan
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Amogh Sirnoorkar, Dean Zollman, James T. Laverty, Alejandra J. Magana, Sanjay Rebello, Lynn A. Bryan
Source Publication: View Original PaperLink opens in a new window
Project Contact: Dr. Jianhua Yang
LLM Model Version: gpt-4o-mini-2024-07-18
Analysis Provider: Openai