Using AI to create a selfie was a revealing experience because it changed how I approached self-representation. Instead of controlling my appearance through a camera, I had to articulate who I am through text-based prompts. This process made me aware that AI does not “see” me as a person, but interprets my identity through language, data, and visual patterns. Writing prompts forced me to reflect on how I understand myself within digital spaces rather than how I look in everyday life.
The prompts I used focused on themes such as digital identity, algorithmic mediation, and cultural visibility. I included references to interface elements, fragmented surfaces, and data-like visuals to guide the AI toward a more conceptual representation. These prompts directly influenced the final image by prioritizing symbolic and abstract elements over realism. As a result, the image emphasizes digital aesthetics and mediated visibility instead of functioning as a conventional photographic selfie.
The AI-generated selfie partially aligns with my self-perception. While it does not accurately represent my physical presence offline, it strongly reflects how I experience myself online. The image captures a sense of being curated, observed, and shaped by digital systems that influence how identity is presented and interpreted. This gap between my offline self and the AI-generated image is significant, as it highlights how self-representation through AI is always mediated by algorithms, datasets, and cultural norms. Rather than seeing the image as a direct portrait, I understand it as a visual interpretation of how my identity is processed and represented within algorithmic environments.
Hello Jalynn, your idea is very interesting. You think that although this picture generated by AI doesn't look like a real person, it is very similar to your online presence. In fact, we describe it using keywords and text, and AI finds pictures with the same definitions in its database and combines them. In reality, this is how AI responds to our instructions using the most old visual patterns it has learned from the internet.
ReplyDeleteSo, that seemingly cool "digital self-portrait" is actually a work created by combining our ideas with the templates in the AI database. All the profound interpretations made by AI after this painting were like giving this idea a reasonable explanation. The explanations were always correct and couldn't be found to be wrong. However, they never met people's requirements and thoughts. So when AI enters human work and provides assistance, increasing efficiency, humans cannot truly control the output results.
Then, the uncontrollable results will affect people's original judgments and thinking? I think this is inevitable. When we give an answer, but others give a different answer, most people will start to doubt whether their answer are correct. These may seem insignificant in small decisions, but when it comes to major decisions, this may lead to irreparable mistakes.
Thank you for sharing Jalynn!
ReplyDeleteYour observation that AI "does not 'see' me as a person, but interprets my identity through language, data, and visual patterns" gets at something fundamental about these systems; they don't represent us, they translate us through their own logic.
I'm particularly interested in your distinction between your "physical presence offline" and how you "experience myself online." This suggests that algorithmic self-representation isn't just about accuracy but about capturing different dimensions of identity that emerge in digital spaces. Your choice to prioritize "symbolic and abstract elements over realism" is an interesting approach that treats the assignment as an opportunity to explore how digital systems mediate identity rather than simply mirror it.
For Assignment 3, I encourage you to build on this strong conceptual foundation by including specific examples of your prompts and analyzing how the AI responded to particular terms. What does it mean to be "curated, observed, and shaped by digital systems"? Whose perspectives and values are encoded in these algorithmic interpretations? You've laid excellent groundwork for exploring these questions in depth.
Excited to see where you take this analysis!