Integrating Artificial Intelligence with Augmented Reality for Scientific Modeling

• In Event: AI as a Co-Pilot in Science Education: Reimagining Instructional Practice

Abstract

Objectives or Purpose
This study investigates the integration of artificial intelligence (AI) with augmented reality (AR) to enhance scientific modeling in high school science classrooms. Our objective is to develop and evaluate an AI-driven instructional agent paired with AR visualizations to foster dynamic, inquiry-based learning experiences that align with Next Generation Science Standards (NGSS) modeling practices. We aimed to deepen students’ conceptual understanding of complex scientific phenomena, such as protein synthesis and lactose intolerance, while promoting equitable, engaging, and personalized instruction.

Theoretical Framework
Our work was grounded in situated learning theory, which posits that learning is most effective when embedded in authentic, interactive contexts (Lave & Wenger, 1991). We extend this framework by leveraging AI and AR to create immersive environments that mirror scientific inquiry processes. The AI facilitates scaffolded learning through real-time dialogue and adaptive feedback, while AR enables students to visualize and manipulate abstract concepts, fostering deeper engagement with scientific models. This approach aligns with NGSS’s emphasis on modeling as a core scientific practice.

Methods, Techniques, or Modes of Inquiry
We developed an AI-driven instructional agent integrated with AR technology to guide small-group science labs. The AI introduces key concepts, monitors student understanding through natural language processing, and transitions students to AR-based modeling activities when they demonstrate readiness (e.g., articulating concepts like “gene expression”). In the AR environment, students manipulate 3D models of scientific phenomena, such as enzyme interactions in lactose digestion. We conducted a quasi-experimental pilot study in high school biology classrooms, comparing the AR-AI hybrid approach to traditional paper-based instruction.

Data Sources or Evidence
The study involved 120 high school students across four biology classrooms. Data sources included pre- and post-assessments of conceptual understanding, classroom observations, student discourse transcripts, and engagement surveys. The AI system logged student interactions, capturing keywords and responses to assess readiness for modeling tasks. Video recordings of AR lab sessions provided qualitative insights into collaborative modeling practices. Comparative data from control groups using traditional lessons were analyzed to evaluate differences in outcomes.

Results and/or Substantiated Conclusions
Preliminary findings indicate that students using the AR-AI system exhibited higher engagement and used more scientific vocabulary compared to the control group. Post-assessments showed improved understanding of protein synthesis and lactose intolerance, with students demonstrating stronger alignment with NGSS modeling practices. Qualitative analysis revealed that the AI’s adaptive feedback and the AR’s immersive visualizations supported collaborative exploration and deeper conceptual connections, particularly for students with diverse learning needs.

Scientific or Scholarly Significance of the Study
This study contributed to the emerging field of technology-enhanced science education by demonstrating how AI and AR can create responsive, equitable learning environments. By scaffolding inquiry and visualizing phenomena, the AR-AI system empowers students to engage in authentic scientific modeling. For teachers, the system serves as a real-time instructional partner, enhancing their ability to deliver high-quality, inclusive science education. These findings inform the design of next-generation learning tools and advance the integration of immersive technologies in NGSS-aligned curricula, contributing to the innovative technologies for equitable STEM education.

Authors

• Bryan A. Brown, Stanford University

• Alan Cheng, Stanford University

• Lisa Teresa Archuleta, Stanford University

• Elizabeth Childs, Stanford University

• Roy D. Pea, Stanford University

• Laurence Tan, University of California - Los Angeles