Skip to main content Skip to navigation

Adaptive Gen-AI Guidance in Virtual Reality: A Multimodal Exploration of Engagement in Neapolitan Pizza-Making

Project Overview

The document explores the integration of generative AI (Gen-AI) in education, particularly through a virtual reality (VR) system for procedural learning exemplified by Neapolitan pizza-making. It emphasizes the role of adaptive Gen-AI in boosting learner engagement by customizing instructional guidance according to individual profiles and facilitating real-time interactions. Key findings from the study reveal that a moderate level of adaptivity enhances learner engagement by effectively directing visual attention and minimizing extraneous exploratory behavior. However, it also cautions that excessive adaptivity can lead to user overwhelm, potentially hindering the learning experience. Overall, the application of Gen-AI in this context demonstrates a promising approach to personalized education, fostering a more engaging and tailored learning environment.

Key Applications

Neapolitan Pizza VR - an immersive VR environment for pizza-making

Context: Educational context focusing on procedural learning of cultural heritage, targeting learners interested in culinary arts and cultural practices.

Implementation: Implemented as a VR simulation using Unity, where an adaptive Gen-AI tutor provides guidance based on user demographics and real-time actions.

Outcomes: Increased user engagement through higher visual attention towards the AI tutor and reduced unnecessary exploratory behavior, with moderate adaptivity yielding optimal engagement.

Challenges: Evaluation of adaptive AI systems is complicated by variability in AI response times; balancing adaptivity levels is crucial to avoid overwhelming learners.

Implementation Barriers

Technical Barrier

Inherent variability in AI-generated responses can introduce inconsistencies in user experience and engagement metrics.

Proposed Solutions: Systematic control of AI response variability, potentially by standardizing response timing or exploring alternative response generation methods.

User Experience Barrier

Excessive adaptivity could overwhelm learners, while minimal adaptivity risks disengagement.

Proposed Solutions: Defining an adaptivity threshold to balance instructional guidance without diminishing learner autonomy.

Project Team

Ka Hei Carrie Lau

Researcher

Sema Sen

Researcher

Philipp Stark

Researcher

Efe Bozkir

Researcher

Enkelejda Kasneci

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Ka Hei Carrie Lau, Sema Sen, Philipp Stark, Efe Bozkir, Enkelejda Kasneci

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