Surprise Machines

Surprise Machines is a visual investigation that will take the form of a digital installation, part of the exhibition Curatorial A(I)gents that took place at the Harvard Art Museums in Spring 2022. The project sets out to visualize and curate the entire universe of museums’ collections, intending to open up unexpected insights into the more than 200,000 objects that make them up. To accomplish these surprise encounters, “black box” algorithms are curatorially employed to shape the data visualization, and a “choreographic interface” has been designed to connect the audience’s movement with several unique views of the objects.

From a technical point of view, Surprise Machines relies on the Harvard Art Museums’ API, which provides access to images and metadata of more than 200,000 objects. Using the PixPlot library from the Yale University DHLAB, the images from the API are arranged in the Cartesian plane using the UMAP algorithm. The interface uses machine vision to mediate the interactive gestures and AI to interpret them in each still frame reviewed in a Python script.

Surprise Machines is part of Curatorial A(i)gents, a group exhibition presenting a series of machine-learning-based experiments with museum collections developed by members and affiliates of metaLAB (at) Harvard.