AI is helpful in expediting repetitive, foundational procedures, however, it is terrible at being used as a high level technological tool to create analytical results.
I find AI very useful for refining wording- making simple, foundational changes to work. For example, if I give AI(using ChatGPT in this case) a rough paragraph outline of what I want to say in terrible grammatical order:
“AI is bad but cool we need to use it with caution and with some professional knowledge in the field you are applying it to. We should explore the ethics of AI and potential ways of guarding the rights of authors to original information”
AI will return something like this back to me:
“AI is both fascinating and potentially problematic, requiring careful and informed use. Its application should be guided by professional expertise in the relevant field to ensure responsible deployment. Additionally, it is crucial to explore the ethical implications of AI, particularly in safeguarding authors’ rights and protecting original content.”
The paragraph is much more refined and it probably just saved me 20 minutes of time and brain power! Obviously the wording doesn’t sound as authentic as human voice but you always go back and make edits or as AI to make more edits for you.
An ethical issue that is potentially posed here is whether or not AI took someone else’s work and directly rephrased it into my words. That’s why having a background in the field you’re working in is crucial. This knowledge becomes particularly vital when writing literary reports, as AI may draw from published papers, raising concerns about plagiarism. Having that foundational knowledge would somewhat allow you to identify potential plagiarism concerns.
Another issue with AI is its tendency to smooth over flaws, rendering the issues almost unrecognizable through its polished language. As Ted Chiang puts it:
You’re still looking at a blurry JPEG, but the blurriness occurs in a way that doesn’t make the picture as a whole look less sharp
Ted Chiang, “ChatGPT Is a Blurry JPEG of the Web”
AI is not precise enough at achieving higher level complex solutions. It draws from existing solutions and attempts to make predicitions through a simplified process to maximize efficiency. This simplified processing is what leads the solutions to be inaccurate and lacking in nuance. It’s like a game of telephone—each time AI processes information, something is inevitably lost in translation. With every step, meaning gets subtly distorted, until the final message is more refined but less faithful to the original.
This issue is also reflected in art restoration:
“These recreations don’t teach us anything we didn’t know about the artists and their methods.”
Sonnja Drimmer, “How AI is hijacking art history”
AI has attempted art restoration, but once again, because it draws from pre-esisting information, it is unable to tell us anything more about the artwork. In addition, due to AI’s simplified computational processing, not only are you not learning more, you’re losing valuable pixels and information in the restored attempts.
4 thoughts on “Color “Restoration” – Lab 4”
Leave a Reply
This site uses Akismet to reduce spam. Learn how your comment data is processed.
One of the other problems with using AI to write paragraphs for yourself is you lose out on the opportunity to improve your own skills and speed. As for colorizing photos, it is interesting that the AI is choosing to make the pictures in the background really vibrant yet the rest of the photo is muted.
Hi Jennifer! I agree with a lot of what you said. I thought you brought up some really interesting points regarding AI, especially your comparison of it to a game of telephone. I completely agree with you that it is very similar to the game, as AI often distorts meaning in subtle ways. AI can be helpful in rephrasing sentences but it often loses the nuanced details that make the original message meaningful.
I really like the the connection you made to a long game of telephone. It is a good way to conceptualize the loss of data over a repeated use of AI output. While it might not happen right away, over time, the level of accuracy and information diminishes through AI. I also agree with pointing out the AI does not produce original output but instead draws from previous solutions. I think this is an important topic regarding AI and as you mentioned leads to quite a few problems.
I agree with all of your points about the drawbacks of AI usage. To add on to the point you made about how AI lacks human voice, I think this also connects to how AI tends to hinder the users’ opportunity to learn and grow. Instead of learning how to do something like code or refining a paragraph users might take the easy route by using AI. This could potentially lead to an increased reliance on the tool rather than fostering our own skills.