Understanding VTuber dynamics when the performer is an AI

Determine how audience engagement patterns, parasocial relationship formation, and monetization practices documented for human VTubers unfold when the performer is a fully autonomous AI VTuber whose dialogue and behavior are generated by large language models, rather than a human performer.

Background

Prior HCI scholarship on VTubers has examined avatar and identity design, the relationships between VTubers and fans through parasocial interaction, and the economic infrastructures of livestreaming communities. These insights, however, are grounded in human-led VTubing and may not directly translate to contexts where the performer is an AI agent.

The paper highlights a gap: while technical feasibility shows AI VTubers can exist, it is unclear how the established dynamics of participation, authenticity, and fan economy behave when the performer is not human. This open question motivates the authors’ case paper of Neuro-sama but remains broader than a single instance.