ethical issue:
I think the most significant problem with AI when dealing with graphics is that it is making “assumptions.” Not that the assumptions are incorrect, but the AI is not making causal inferences behind its assumptions; it’s relying on the data to make judgments. It’s true that data can tell us a lot, and we can make a lot of logical judgments based on data. However, the problem is that it doesn’t always work that way. Bias exists in such a process in two ways. Firstly, there may be bias in the algorithm of AI itself, so unfair or discriminatory outcomes may be produced. On the other hand, bias may exist in the training data. Sometimes the amount of data doesn’t tell us much about its reliability but rather is misleading. Also, the data may fail to represent the entire group.
Some point out that privacy and security may become concerns. AI imaging tools may inadvertently reveal sensitive information, and the potential misuse of such data may raise significant privacy concerns. Also, AI systems used in graphing and imaging may be vulnerable to attacks, leading to data leakage or manipulation.
Another ethical issue worth discussing is the value behind such AI. Are they genuinely contributing to the artistic and historical value with those images they manipulated? This is a good question I have from the reading. I think on the artistic side, there is a concern about the actual artistic intent of the work being distorted or lost. On the historical side, AI tools may create highly realistic work that later becomes very difficult to distinguish from actual historical artifacts. This raises concerns about the potential for deception.
quotes:
But outside of flexing the prowess of AI, is there any value – artistically, historically – to what the company is doing?
Sonja Drimmer, Associate Professor of Medieval Art, UMass Amherst
By diving deep behind ai company, Oxia Palus, Sonija noticed multiple points worth discussing.
theoretical problems are steamrollered flat by the weight of data
Art historian Claire Bishop, Against Digital Art History
original and colorized images
2 thoughts on “reflective blog week 4”
Leave a Reply
This site uses Akismet to reduce spam. Learn how your comment data is processed.
Hi Emma! I think you bring up a great point about the negatives of using only data averages and making assumptions. I think this is pretty common, but AI uses this data processing as its main way to return answers, and this is a big issue because it ignores outliers, which are crucial to look at and understand. I also never thought about how colorizing images could actually be an issue about personal privacy. Great post!
Hi Emma, your observations on AI assumptions in graphics resonate. The dual bias sources, algorithmic and data-related, are critical challenges. Privacy and security concerns underscore the need for robust safeguards. The ethical question on the artistic and historical value of AI-manipulated images prompts essential discussions. Addressing these issues demands ongoing interdisciplinary dialogue for responsible AI development.