Sanyu Studio: A Multi-Agent System for Art-Historical Narrative Construction

A multi-agent system named Sanyu Studio has been developed to explore how LLM-based interaction can support plural art-historical narrative construction. This system models 321 Sanyu oil paintings as agents, incorporating factual, interpretive, organizational, and memory-filtering mechanisms. A workshop with art-university participants demonstrated that user prompts, evidence organization, and cognitive tendencies contribute to divergent yet coherent interpretations, suggesting AI's potential to amplify human agency in art-historical interpretation, especially where historical evidence is limited.

Why it matters

This development challenges the concern that generative AI may standardize interpretation by demonstrating its capacity to foster diverse narratives. It highlights the potential for AI tools to enhance human agency in complex interpretive domains, providing new methodologies for research and public engagement. This approach could significantly impact how knowledge is constructed and disseminated in fields reliant on expert interpretation and limited historical data.

Key insights

• Sanyu Studio is a multi-agent dialogue system designed to facilitate plural art-historical narrative construction.
• The system models 321 Sanyu oil paintings as agents, each equipped with fact, interpretation, organization, and memory-filtering mechanisms.
• A seven-day workshop involving eight art-university participants revealed that user prompts, evidence organization, and individual cognitive tendencies influenced the creation of diverse yet coherent digital Sanyu interpretations.
• The research suggests that AI can amplify human agency in art interpretation, particularly in contexts with limited historical evidence.
• The system offers an interactive entry point for public audiences into art-historical interpretation.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2608.18677