Original (top), AI-Colorized (middle), and Distant Viewing Explorer (bottom) versions of the same photo show how dramatically perception shifts depending on the way an image is processed.
When I first colorized this old photograph with an AI tool, I was surprised — not by the colors themselves, but by how convincing they felt. Once I saw the colorized version next to the original black-and-white image above, I almost believed I was looking at how the scene actually looked in real life. But that’s exactly the ethical issue here: AI doesn’t reveal the past, it interprets it. And those interpretations can shape how we remember history. Sonja Drimmer points out that colorization tools don’t really “bring events back to life,” they “recreate what is already a recreation — a photograph — in our own image, now with computer science’s seal of approval.” Drimmer’s observation reminds me that we’re not uncovering hidden truth, we’re layering modern assumptions onto already mediated images (Drimmer).
That’s important because once something is in color, it feels authoritative. We tend to trust colorized photos as more real or accurate, even when the algorithm is just guessing. AI has no firsthand knowledge of what the colors would have been, it just predicts tones based on patterns in its training data. The Distant Viewing Explorer output makes this even clearer: its object labels and visual detections are often inaccurate or overly simplified, identifying shapes in ways that don’t really match what a human viewer would see. And as Drimmer warns, these tools can easily blur the line between representation and reality if we’re not careful about context. Another layer to this issue is one that’s less visible but equally important: where the AI learned to colorize in the first place. Most of these models are trained on massive datasets that include works created by real artists and photographers. When those works are used without consent or attribution, it raises questions about intellectual property and artistic ownership.
Works Cited
Drimmer, Sonja. “How AI Is Hijacking Art History.” The Conversation, https://theconversation.com/how-ai-is-hijacking-art-history-170691. Accessed 30 Jan. 2026.
This comparison is such a cool way to visualize the issue! I love how you connected the visual convincingness of the colorized photo to the danger of blindly trusting AI. It really drives home the point that we see a modern guess rather than actual history. It is fascinating and a little worrying how easily color tricks our brains into thinking this is the truth. You also raised a great point about the training data ethics at the end. It makes me wonder how much of our restored history is actually just stolen patterns from other artists.