Researchers have developed a multi-agent system called Sanyu Studio to explore how generative AI can be used in art-historical narrative construction. The system uses large language models (LLMs) to interact with users and model the creation of digital versions of oil paintings by Chinese artist Sanyu. A seven-day workshop with art university participants showed that user prompts, evidence organization, and cognitive tendencies influenced the creation of divergent yet coherent narratives. The study suggests that AI can amplify human agency in art-historical interpretation when faced with limited historical evidence. --- Why it matters: This research matters to engineers and researchers in AI because it explores how generative models can be used to support plural and interactive art-historical narrative construction, which has implications for the role of AI in cultural heritage preservation and public engagement. Source: https://arxiv.org/abs/2608.18677