The Limits of Colorization of Historical Images by AI

Contrary to what you might think, a problem isn't occurring because there are more white people in some historical photo dataset.

For most of the history of photography, if you had a black and white photograph and wanted to see it in color, you had to add the pigment with a paintbrush. Recently though, digital methods, and now artificial intelligence-driven methods, have made colorization available to everyone. This is a victory for technological progress! Or is it? As a recent twitter thread demonstrated, AI colorization algorithms don’t quite get it right. If you take an existing color image, convert it to grayscale and use an AI colorization algorithm to repaint it, the re-colorized image looks dull in comparison. To

related reading

• The Controversial History of Colorizing Black-and-White Photos | PetaPixelpetapixel.com
• Are Colorized Photos Rewriting History?gizmodo.com
• Using Artificial Intelligence to Augment Human Intelligencedistill.pub
• A Group of Historians Is Arguing to End the Coloring and Modernization of Old Historic Photos | Fstoppersfstoppers.com
• Touching up: are we wrong to colourise old photographs? – The Irish Timesirishtimes.com
• The Passionate Photo Colorizers Who Are Humanizing the Pastatlasobscura.com
• AI photo restoration shines a light on life in old Ireland | CNNcnn.com
• Epistemic calibration and searching the space of truththesephist.com
• Artificial Intelligence Has a Racial and Gender Bias Problemtime.com
• The Illustrated Stable Diffusion – Jay Alammar – Visualizing machine learning one concept at a time.jalammar.github.io
• Why AI alignment could be hard with modern deep learningcold-takes.com
• How Did You Do On The AI Art Turing Test?astralcodexten.com