Looking Ahead to Look at the Past: Using Artificial Intelligence to Re-Analyze 19th-Century Photographs

Talk | Auditorium | 2p

Reception | Terrace Room | 3p

Like most then-popular composite group class portraits, the image personifying all 247 students of the Harvard University Class of 1887 was created by practitioners in a commercial studio. To make it, an unnamed photographer at Lovell Photo sequentially exposed the members’ portraits on a light-sensitive plate after meticulously aligning and resizing each negative by the axis of their eyes and the center of the face. The studio lucratively sold these as keepsakes to members of the class, their families, and to universities.

Artificial Intelligence and facial recognition suggest that the mathematical likelihood of inclusion of each the portraits of all 247 class members varies significantly. The author of this image chose certain images over others, gave a few of them greater prominence, and omitted many of the classmates altogether. Although we will never know why the maker of this composite felt biased toward inclusion or exclusion of certain images, A.I. allows us to rediscover those decisions and speculate, with the benefit of the image’s commercial socio-economic contexts, about the reasons for doing so. This lecture features such an interdisciplinary history.