I went into this lab assignment thinking it would be something easy and purely technical… yet it turned out that there is more into it...
It took me a while to figure out how to grab the IIIF link from CONTENTdm, changed the URL to full/full/0/default.jpg (thanks, IIIF). Once that part was figured out, running it through DeOldify was fairly easy.
I can’t deny that this AI recoloring tool was a great way to attempt to see see what was the past like, but I can’t stop thinking about the ethical struggle behind it. One biggest thing is how AI could affect people’s opinions on truth. Whatever AI is doing to the black-and-white picture wasn’t time-traveling. It is just AI’s best guess based on the information we provided it which connects back to Sonja Drimmer’s argument in How AI is Hijacking Art History, she puts, “They have not, in actuality, revealed one secret or solved a single mystery. What they have done is generate feel-good stories about AI”. When my output looks “real,” it’s tempting to treat it like recovered truth. For But again, AI dont reaveal any truth even though they could be really helpful and tempting.
AI tools are only helpful when we use it right, so we should be really careful with it. when we use it right, say, in the following pictures, the recoloring helps me noticd a lot of details that I didn’t noticed in the original picture. But I tried to keep in mind that this is AI generated and not truth so i won’t make interpretations. On top of that, I think recognizing a picture is Ai generated would be really helpful for whoever’s looking at it. The key is context: keep the original visible, explain what’s speculative, and link sources so viewers can check for themselves.
2 thoughts on “Recoloring the Past”
Leave a Reply
This site uses Akismet to reduce spam. Learn how your comment data is processed.
I also struggled with the first half of this assignment. Obtaining the IIIF link was definitely a process and I would not have been able to figure out how to do it without Austin’s directions. However, as you mention, once we have that link, running it through DeOldify is alarmingly easy. Far too easy to produce an output that is just a recreation. I agree with you that context is key and as long as viewers know that colorized images are in fact AI generated, it mitigates much of the unethical risk.
Hi Harrah,
I also had a small hiccup with the same portion you had an issue with, because I think I didn’t know which URL to put into IIIF. But I’m glad you were able to resolve it. I really enjoyed the quote from Sonja Drimmer, because I think it resonates well with your ethical issue about truth. I wrote about this idea in my post, so I agree when you noted that the AI doesn’t know what is real and what isn’t. Additionally, I liked that you pointed out a user needs to be cautious when working with AI.