User Acceptance and Concerns for LLM-Powered Conversational Agents in XR

Determine how users’ technology acceptance and concerns are shaped when large language model-powered conversational agents are embedded into extended reality systems and devices, including mixed reality and virtual reality head-mounted displays, in order to understand adoption dynamics and apprehensions specific to this integration.

Background

The paper reviews extensive prior work on user attitudes toward head-mounted displays and conversational agents separately, noting that XR devices collect rich sensor data and that LLM-powered agents enable naturalistic conversations that may prompt sensitive disclosures. This convergence raises distinct privacy, security, social, and trust considerations.

Despite these developments, the authors highlight a gap: it has not been established how users’ acceptance and concerns are specifically shaped when generative-AI/LLM conversational agents are embedded within XR settings and devices. The study in this paper offers baseline evidence via a large-scale crowdsourcing survey, underscoring the need to systematically characterize these perceptions in the XR-LLM context.