1 free read left. Create a free account for unlimited reading + AI research.
Create free accountWhitepaper
DiGRA
The paper argues that meaningful engagement with digital historical figures depends more on structural invitations than on the depth of an AI's factual knowledge.
This paper proposes "Narrative Scaffolding," an interactive narrative strategy for digital humans in museum settings, arguing that meaningful engagement with digital historical figures depends more on structural invitations than on the AI's factual knowledge. Taking a Research through Design approach, the author built a virtual reality prototype featuring a large language model-driven avatar of a female factory worker from 1930s Shanghai, a major industrial center during China's Republican Era. Before interacting, users select a social role (supervisor, coworker, or relative), a choice that sets the context and tone of the dialogue and is framed as a metacommunicative invitation that moves users from observers to participants.
The analysis draws on Celia Pearce's concept of "playframes," the implicit social rules governing interaction, to explain how each role establishes a distinct frame within which user and AI jointly construct a shared experience. This shared experience unfolds in a liminal space between the user's present reality and the historical simulation. Building on Janet Murray's theory of narrative agency, the author positions the user as a "co-performer" of history, confronting them with the ethical and emotional tensions inherent in each role.
The work's central claim is that the design of invitational frameworks, rather than the depth of an AI's historical knowledge, is the primary determinant of whether users actively co-create meaning with digital historical figures. The contribution lies in reframing avatar design in cultural heritage contexts away from the prevailing guide-and-seeker model toward performative, role-based engagement. The paper's evidence base is a single prototype and conceptual argument, so its claims about user outcomes are theoretical rather than empirically measured across a broad sample.
DiGRA · 2026