My Omeka item is linked here or click on the image!
One ethical issue that arrises from AI in image restoration is accuracy of AI in coloring of images/restoration. How do we know AI will properly handle images of POC’s and other issues when it encounters something that it could not be properly trained on? We need to have humans as a check on the accuracy of AI and making sure that it accuratly displays the diversity of people and their cultures.
These two quotes sum up my thoughts pretty well:
But my lingering suspicion emerges from an awareness of how public support for the sciences and disparagement of the humanities means that, in the endeavor to gain funding and acceptance, the humanities will lose what makes them vital.
Sonja Drimmer
Art and history becoming toys in the sandboxes of scientists with no training in the humanities.
Sonja Drimmer
These two quotes sum up my views because AI is built on models of probability and data, but how can we reduce human culture and diversity to a bunch of numbers and models? Human culture is far more intricate and nuanced than what AI can handle, and believing that AI can properly restore or color images accurately is idiotic. While Sonja’s quotes focus mostly on art, I think they’re just as applicable to image analysis.
5 thoughts on “Lab 4 – IMAGE ANALYSIS AND ARTIFICIAL INTELLIGENCE”
Leave a Reply to Lucas Cancel reply
This site uses Akismet to reduce spam. Learn how your comment data is processed.
When it comes to the accuracy of coloring/restoring images, I believe the point you brought up about AI being able to correctly handle images of people with different skin tones is important. At what point could we possibly be changing history by accident with the use of AI restoration on historical images?
I heavily agree with your point that human culture is more intricate than what AI can currently handle. Although these tools are neat and fun to play around with, we need to remember that these models are trained on data that can easily create unreliable reproduction of images. AI will make choices just based on what it has previously seen, which will never fully encapsulate all of human culture.
I totally agree with you! Culture is the thing that a nation is most proud of, and it is far more complicated and rich in content than simple numbers. What matters is all the aspects of people’s lives that cannot be quantified. Reducing cultures to numbers and data is a disrespect of culture. From this perspective, I believe the right way to use AI-generated photo is treating it as an assistance when explaining a nation’s culture to others to make the explanation more vivid , but it should never be used as any definitive evidence for historical investigation.
I agree with your point that we need humans to check the accuracy of AI. While some AI-generated images may be fairly accurate, mistakes are bound to occur. It is an exciting tool to use and fun to play around with, however, in some situations, using AI to replicate images will change the story that is being told. AI is trained on existing information, and I agree with you that thinking AI can 100% accurately reproduce color in images is idiotic.
I like how you talk about the fact that human culture and diversity can’t be reduced solely to numbers. I think that is something very important to consider when working with AI because while AI may be good at compiling numbers and models, it is not capable of accurately replicating the part of us that deals in culture and diversity. I believe that while AI could possibly be useful in image analysis and creation it is important to add a label to those images saying that AI is used so that the audience doesn’t assume them to be completely true and accurate.