The Curator's Machine uses machine learning to support the curation of art collections and exhibitions in museums. The application does not follow the indexed and rather static search logic of previous archives and databases, but uses a special interface to make exploration of mass digitised collections faster and more convenient. Pattern recognition can be used to reveal correlations and connections between works of art that are either difficult for humans to perceive or only partially understood. The strengths of AI become apparent when search results in large image data collections do not remain generically the same. Instead, they adapt based on usage and changing parameters and adapt to the personal interests of the curators.