Google DeepMind: "Aeneas"

Submission for Google DeepMind to the prestigious scientific journal 'Nature,' and accompanying animation. This artwork was developed in close collaboration with the Creative Director and the researchers responsible for the Aeneas project, the first-of-its-kind AI model designed to interpret ancient fragmentary texts. The model was trained on Latin inscriptions from the Roman Empire, and this became the theme of the artwork.

It was important to have a strong and relatable visual narrative to show how the model can be used and we achieved this by symbolically representing the fragmented data sources as dioramas with sections of inscribed text across the face of the plinths. The inscription we used is a real example of one of the texts restored using Aeneas, and the dioramas depict snapshots of various geographic and historical scenes that would have been commonplace during this era, illustrating the ability of Aeneas to contextualise and connect relics from all over the world.