SimPal: Towards a Meta-Conversational Framework to Understand Teacher's Instructional Goals for K-12 Physics
Project Overview
The document explores the utilization of generative AI in education through the development of SimPal, a meta-conversational AI framework tailored for K-12 science teachers to enhance the inclusion of physics simulations in their teaching strategies. By addressing the gap between static instructional goals of traditional AI agents and the dynamic pedagogical requirements of educators, SimPal empowers teachers to express their instructional aims via natural conversation. Leveraging advanced large language models, such as ChatGPT-3.5 and PaLM 2, the framework effectively interprets teachers' inputs to identify relevant physical variables, facilitating a more personalized and adaptable use of AI in the classroom. This innovative approach not only promotes better alignment with teachers' needs but also enhances the overall educational experience, showcasing the potential of generative AI to transform teaching methodologies and improve student engagement in science education.
Key Applications
SimPal - a meta-conversational agent for physics simulations
Context: K-12 education, specifically aimed at science teachers using physics simulations.
Implementation: Teachers engage in natural conversations with SimPal to articulate their instructional goals, which SimPal then uses to identify relevant physical variables for simulations.
Outcomes: SimPal improves the alignment of AI agents with teachers' instructional goals, enhancing the effectiveness of simulation-based learning in physics.
Challenges: Teachers often lack technical expertise to customize AI agents, leading to reliance on pre-existing solutions that may not meet their specific needs.
Implementation Barriers
Technical and Adoption Barrier
Teachers often lack the technical expertise to customize AI agents, which can limit the effectiveness of AI in educational settings. Additionally, teachers are hesitant to adopt new third-party simulations due to concerns about alignment with their instructional goals and the time required to learn new tools.
Proposed Solutions: SimPal allows teachers to customize AI agents without needing deep technical knowledge, facilitating easier integration of AI into their instructional practices. It also enables teachers to freely design their own instructional goals and quickly adapt existing simulations to meet those goals.
Project Team
Effat Farhana
Researcher
Souvika Sarkar
Researcher
Ralph Knipper
Researcher
Indrani Dey
Researcher
Hari Narayanan
Researcher
Sadhana Puntambekar
Researcher
Shubhra Kanti Karmaker
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Effat Farhana, Souvika Sarkar, Ralph Knipper, Indrani Dey, Hari Narayanan, Sadhana Puntambekar, Shubhra Kanti Karmaker
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