Integrating Personality into Digital Humans: A Review of LLM-Driven Approaches for Virtual Reality

Researchers have published a comprehensive review of methods for integrating large language models (LLMs) into virtual reality environments to create more realistic digital humans with personality traits. The study explores various approaches including zero-shot, few-shot, and fine-tuning methods while highlighting challenges like computational demands and latency issues that need to be addressed for practical applications.

• →LLMs can be integrated with multimodal outputs like facial expressions and gestures to create personality-driven virtual agents in VR environments.
• →The research explores multiple approaches including zero-shot, few-shot, and fine-tuning methods for implementing personality traits in digital humans.
• →Key challenges include computational demands, latency issues, and lack of standardized evaluation frameworks for multimodal interactions.
• →Applications span across education, therapy, and gaming sectors with potential to transform human-computer interaction.
• →The work aims to foster interdisciplinary collaboration to advance VR-based digital human experiences.