Installation Personas: A Persona-Centered Framework for AI-Driven Interactive Art Installations

Abstract

Artificial intelligence (AI) is enabling interactive art installations to become increasingly adaptive and socially interactive, yet limited conceptual guidance exists for analyzing installations themselves as persona-like entities. In this study, the Installation Persona Framework is proposed, extending persona-centered analysis from users and individual AI agents to AI-driven interactive installations. The framework conceptualizes Persona Identity through two complementary dimensions—Persona Type and Persona Role—and examines their expression through Persona Behavior and Interaction and association with Artistic Function and Intended Audience Experience. Using a structured screening process, 41 AI-driven interactive installations were selected from 329 artworks presented at the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) AI Art Gallery from 2024 to 2026. Qualitative analysis identified seven recurring Persona Archetypes and eight higher-order Persona Role Families. An independent classification assessment carried out by four external coders showed statistically significant but fair intercoder agreement for Persona Archetype (Fleiss’ κ = 0.266) and Persona Role Family (κ = 0.203), indicating measurable but limited classification reproducibility and ambiguity in category boundaries. Because audience evaluation was not conducted, audience experience was limited to artist-intended or researcher-inferred interpretations based on available documentation. Overall, the framework provides a structured vocabulary for analyzing AI-driven interactive installations as integrated persona-like entities.

The authors' abstract, as published at the source. Applied Sciences, 2026 · DOI ↗