In this week’s lab, I used a Google Colab AI model to colorize a black-and-white photograph. The results clearly show the potential AI has, but they also raise ethical questions about how such tools shape what viewers take to be “real.” For this assignment, I chose the photograph of the chapel interior from the choir loft from the Carleton digital archive, and I have posted the colorized version on Omeka. Below are the original images with its colorized version.
As the quote below suggests, a key concern is that colorization can project a presumed reality onto an image made in a very different context. Without a clear note that the picture was processed by AI, viewers may assume the colors reflect the “authentic” scene at the time of capture. In practice, color choices, whether introduced by model assumptions, template biases, or rendering settings, are interpretive and manipulable. As a result, AI-colorized images risk fabricating aspects of the original and misleading the audience.
How absurd to think that black-and-white photographs from 100 years ago would produce colors in the same way that digital photographs do now.
Sonja Drimmer, How AI is Hijacking Art History, The Conversation
Some argue that AI-generated images and other outputs, though inauthentic, can still enrich our understanding of distant artifacts. I disagree with the proposed claim. As the quote below suggests, these renderings do not contribute verifiable information; they contribute interpretation. Any “new” detail arises from model assumptions, training-data biases, and rendering choices, not historical fact. The task is to separate bias from knowledge. Therefore, AI renderings should be clearly labeled, paired with a disclaimer and context notes, and treated as hypotheses or illustrations, not evidence, unless corroborated by independent sources.
These recreations don’t teach us anything we didn’t know about the artists and their methods.
Sonja Drimmer, How AI is Hijacking Art History, The Conversation
