Skip to main content Skip to navigation

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

Let us know you agree to cookies