Museums and cultural institutions increasingly use artificial intelligence (AI) to manage collections, interpret content, and engage audiences. Most existing guidance, such as ethics toolkits and governance models, supports planning before implementation. Less attention has been given to AI applications already in operation. This work proposes a framework for assessing existing AI applications in cultural institutions, examining not only what these systems do, but how they are developed, governed, and sustained. Drawing on human-computer interaction (HCI), cultural policy, and museum studies, the framework has five dimensions. These are strategic alignment, ethical design, participatory development, institutional readiness, and iterative evaluation. Each dimension is operationalized through a guiding question and an interpretive rating scale. The framework is applied to four international AI applications through qualitative analysis of publicly available documentation. The cases are the Hello Vincent conversational installation at the Musée d’Orsay, Smartify personalized tours at the Smithsonian American Art Museum, the exhibition Edvard Munch: Connect the Lines at the Munch Museum, which uses AI to match visitors’ drawings with the collection, and machine learning for audiovisual cataloguing at the National Film and Sound Archive of Australia. Findings show that strategic alignment with institutional missions is consistently strong. Ethical design is addressed unevenly, with one case offering a model of transparency and consent-based data governance. Participatory development remains limited, and evaluation practices are limited or only partly documented. The framework is intended as a reflective tool that helps institutions examine and improve how they develop, govern, and evaluate their AI applications, rather than as a performance metric. Future work will extend the framework with more case studies and a wider range of AI applications, refining it in collaboration with museum professionals and other stakeholders. The poster invites discussion on how HCI methods can support responsible and mission-aligned AI in the cultural sector.
We are excited to invite you to the ACM GreekCHI-CYCHI Symposium 2026, held on November 6-7, 2026 jointly in Athens, Thessaloniki, Nicosia and Online.