Towards Embodied Conversational Agents for Reducing Oral Exam Anxiety in Extended Reality

2025-08-15

AI Summary

Oral examination anxiety significantly impairs undergraduate cognitive performance and academic development. This paper proposes a novel oral examination training system integrating Extended Reality (XR), Embodied Conversational Agents (ECAs), and real-time Large Language Models (LLMs). The system delivers a highly immersive, repeatable virtual examiner environment that ensures emotional safety, provides personalized feedback, and supports adaptive spoken responses. Its core innovation lies in the tight coupling of ECAs and LLMs: examiners exhibit human-like multimodal behaviors while dynamically generating domain-accurate, assessment-aligned follow-up questions and feedback grounded in students’ real-time verbal responses. A prototype implementation demonstrates feasibility in reducing anxiety and enhancing self-efficacy. The work establishes a scalable, empirically grounded paradigm for educational psychological intervention and contributes a foundational technical architecture and design principles for XR–AI–enabled pedagogical applications.

Technology Category

• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Machine Learning: Large Multimodal Models (LMMs)
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Agentic search
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Oral examinations are a prevalent but psychologically demanding form of assessment in higher education. Many students experience intense anxiety, which can impair cognitive performance and hinder academic success. This position paper explores the potential of embodied conversational agents (ECAs) in extended reality (XR) environments to support students preparing for oral exams. We propose a system concept that integrates photorealistic ECAs with real-time capable large language models (LLMs) to enable psychologically safe, adaptive, and repeatable rehearsal of oral examination scenarios. We also discuss the potential benefits and challenges of such an envisioned system.

Problem

• Reducing oral exam anxiety using embodied conversational agents
• Integrating ECAs with LLMs for realistic exam rehearsal
• Exploring XR environments for adaptive oral exam preparation

Innovation

• Photorealistic ECAs for exam rehearsal
• Real-time LLMs for adaptive interactions
• XR environments for safe practice