My Favorite Streamer is an LLM: Discovering, Bonding, and Co-Creating in AI VTuber Fandom
This is a strong CHI qualitative paper with a clear and timely contribution: it shows that AI VTuber fandom is not a novelty wrapper around human-fandom theory, but a setting where authenticity, attachment, and financial support are reorganized around consistency and co-performance. The empirical base is solid for a single-case study, though generalization should stay cautious.
Axes Lens
Rare contribution shape, typical evidence profile. The point here is not a score. It is to show what kind of claim the paper makes, and whether the evidence pattern is unusual or baseline in this 268 -review set.
Contribution shape
- Knowledge form
- descriptive knowledge typical · 92/268
- Novelty type
- theory typical · 15/268
- Abstraction level
- field typical · 41/268
- Generalization target
- field argument typical · 55/268
- Validation mode
- mixed methods typical · 136/268
Evidence profile
- Evidence strength
- strong typical · 158/268
- Claim alignment
- strong typical · 231/268
- Overclaim risk
- medium typical · 210/268
Review Summary
This paper earns its honorable-mention status by making a genuinely interesting move in the VTuber/fandom literature: it takes an object that is explicitly non-human and shows that audiences still build recognizable forms of attachment, authenticity judgments, and participatory labor around it. The central theoretical shift is persuasive because the authors do not simply claim that AI VTubers are “like” human VTubers; instead, they argue that consistency of persona can stand in for humanness as an authenticity anchor, and that financial support can function as a mechanism for steering live content rather than only rewarding performance. That is a useful reframing for CHI because it connects fandom practice to AI-mediated interaction design and community governance. The validation is appropriate for the claim type: a mixed qualitative design with a large survey, interviews, and interaction-log analysis gives enough triangulation to support the descriptive and theoretical account. At the same time, the paper is careful enough to acknowledge that it is a single-case study centered on Neuro-sama, with recruitment from fan communities, English-language Twitch logs, and no equivalent qualitative comparison from human VTuber audiences for some research questions. Those limits matter, but they do not undercut the contribution; they mainly constrain how far the authors can generalize. Overall, the paper’s novelty is in the conceptual vocabulary it offers and in the way it repositions AI fandom as active co-creation rather than passive consumption. The main caution is to avoid reading the findings as universal across all AI creators or all fandoms; they are strongest as a field-level argument about what becomes possible when the performer is algorithmically generated yet socially legible as a stable persona.
What Changed
Canon before
Prior CHI work on VTubers and fandom has largely centered human performers, parasocial interaction, authenticity, and community economies; this paper shifts the lens to an explicitly non-human AI VTuber and asks how fandom forms when the performer is algorithmically generated.
Departure from common sense
The paper argues that audiences can form parasocial intimacy and treat authenticity as meaningful even when the streamer is explicitly non-human. That runs against the common expectation that fandom attachment depends on a human performer, backstage labor, or at least a human “real person” behind the avatar.
Actual novelty
The paper’s main novelty is a qualitative account of AI VTuber fandom that reframes authenticity around persona consistency and financial support around participatory co-performance. It names these as “consistency-as-authenticity” and “real-time co-performance commodification,” extending parasocial and participatory-culture theory to an AI performer context.
Evidence
The study combines a survey of 334 Neuro-sama fans, 12 semi-structured interviews, and livestream interaction log analysis with human VTuber comparison cases. The evidence supports a qualitative account of how fans discover, bond with, and co-create around an AI VTuber, while also showing that the authors explicitly frame the work as theory extension rather than a purely descriptive case report.
“ 2024. VTubing and Its Potential for the Streaming and Design Community: An Austrian Perspective. In Social Computing and Social Media , Adela Coman and Simona Vasilache (Eds.). Springer Nature Switzerland, Cham, 222–233. Digital Library Google Scholar [55] Arleen Salles, Kathinka Evers, and Michele Farisco. 2020. Anthropomorphism in AI. AJOB neuroscience 11, 2 (2020), 88–”
actual novelty · Abstract/Contributions + Discussion 5.1-5.2 · confidence 0.74
“ What remains missing is an account of fully autonomous, LLM-driven entertainers: why humans are drawn to them, how parasocial bonds and community identities form when the agent is visibly non-human, and what economic logics stabilize around such engagements”
departure from common sense · Abstract/Introduction + Discussion 5.2 (Transparent Parasocial Relationships) · confidence 0.78
“ The Research on Applying Artificial Intelligence Technology to Virtual YouTuber. In 2021 IEEE International Conference on Robotics, Automation and Artificial Intelligence (RAAI) . 10–14. Crossref Google Scholar [76] Rui Yan, Zhen Tang, and Dewen Liu. 2025. Can virtual streamers replace human streamers? The interactive effect of streamer type and product type on purchase intention”
limitation · Discussion 5.4 Limitations and Future Work · confidence 0.82
“ We first conducted a survey with 334 Neuro-sama fans to capture broad motivations and engagement patterns, followed by semi-structured interviews with 12 dedicated fans to probe parasocial bonds and community meanings, and finally a log analysis of livestream interaction to examine co-creative practices and financial support behaviors”
validation scope · Method (3 Method) + Interaction log analysis (3.3) · confidence 0.74
Limits
Method limits
The evidence is strongest for a single-case qualitative account. The authors note selection bias from recruiting in fan communities, the absence of equivalent qualitative data from human VTuber audiences for RQ1/RQ2, and the limits of LLM-assisted coding for short messages.
Deployment limits
The findings are most directly applicable to highly engaged fans of a prominent English-language AI VTuber on Twitch. Transfer to other AI creators, other platforms, other languages, or less active audiences is uncertain.
Boundary conditions
The account is bounded by a single prominent case, Neuro-sama, and by English-language Twitch interaction logs. The paper’s claims about authenticity, parasocial bonding, and co-creation are therefore best read as context-sensitive to AI VTuber fandom rather than universal properties of all AI-mediated communities.
Position in field
This paper sits at the intersection of fandom studies, parasocial interaction, and HCI work on AI-mediated communities. Its contribution is to move beyond human VTubers and show how an explicitly non-human performer can still support recognizable fan practices while changing the meaning of authenticity and support.