AI-Powered VR for Anatomy Education: Immersive Learning with Virtual Assistants

Virtual reality (VR) is revolutionizing education, particularly in complex subjects like anatomy. By offering interactive 3D visualizations, VR overcomes the limitations of traditional methods. This article explores the integration of generative artificial intelligence (AI) with VR to create immersive anatomy education experiences featuring AI-powered virtual assistants. These assistants aim to provide personalized support and enhance learning outcomes for students.

Generative AI, exemplified by models like ChatGPT, offers significant advantages over conventional virtual assistants. Unlike rule-based systems, generative AI can produce natural and engaging dialogues, mimicking human-to-human interactions. This technology leverages vast datasets to provide accurate, context-aware, and personalized responses, adapting to the user's level and interests. In anatomy education, generative AI can answer complex questions, provide detailed explanations, and offer customized learning paths.

The study introduces an immersive VR environment designed to support human anatomy education through generative AI conversational assistance. The environment integrates ChatGPT-3.5 to provide intelligent responses and features an embodied virtual avatar with lip synchronization for enhanced realism. A novel framework facilitates communication between the user and the virtual assistant, creating a more interactive and adaptive learning experience.

A pilot user study was conducted with 16 participants to assess the feasibility and effectiveness of the proposed VR environment. The study compared two virtual assistant configurations: avatar-based and screen-based, across two levels of cognitive complexity: knowledge-based and analysis-based questions. The goal was to evaluate user performance, usability, task load, and sense of presence in each configuration.

The results of the pilot study revealed a significant difference in scores obtained from knowledge- and analysis-based questions, particularly in relation to the avatar configuration. This suggests that the presence of an embodied avatar can influence user performance, especially when dealing with more complex cognitive tasks. The study also provided valuable insights into the usability of the VR environment and the effectiveness of the AI-powered virtual assistants.

In addition to performance metrics, the study examined subjective measures such as usability, cognitive task load, and the sense of presence. Participants provided feedback on their experience using the VR environment and interacting with the virtual assistants. This qualitative data helped to identify potential areas for improvement and provided a deeper understanding of the user experience.

The findings of this study have significant implications for medical education. The integration of generative AI into VR environments can create more engaging, personalized, and effective learning experiences for students. Furthermore, the use of embodied virtual assistants can enhance user confidence and participation. The potential benefits extend beyond medical education, with applications in various fields where customized virtual conversational assistants can improve learning and training outcomes.

Future research should focus on refining the VR environment, improving the AI-powered virtual assistants, and conducting larger-scale studies to validate the findings. Exploring different avatar designs, incorporating more advanced AI models, and investigating the long-term impact on learning outcomes are all promising avenues for future research. The ultimate goal is to create a seamless and intuitive learning experience that empowers students to master complex subjects like anatomy with greater ease and confidence. The use of VR and AI in education promises a future where learning is more accessible, engaging, and effective for all.