 |
| AI generated image via Artbreeder |
Creating an AI selfie was an exercise in patience and brutal honesty with oneself. When I first started this assignment, my prompts were a basic description of my physical appearance: “photo of a middle-age, Indigenous woman; short greying blond hair, hazel eyes, black eyeglasses.” And after generating a few images, I found that the AI seemed to get stuck on the “Indigenous” part because it kept generating images with stereotypical Indigenous features: dark hair (despite my hair prompt), dark eyes (despite my eye prompt), and honeyed complexion; it even completed the image with Pendleton-patterned clothing. But when I started refining the prompts, and getting more descriptive of my appearance, it began to generate images that focused more on my age, and what it thought a middle-aged woman was supposed to look like, with greying hair, bifocals, and deep facial wrinkles.
The one thing I noticed though, after several more prompts, was that each new generated image seemed to lose the Indigenous aspect of my identity. For me, my Indigeneity is a major part of how I represent myself, it defines who I am and where I come from. To see it be minimized kind of bothered me to be honest. For this final image, I chose to prompt not just about appearance, but about feeling as well (happy, peaceful, strong). Does this image reflect how I see myself? Not physically (not even close), but there are shades of myself in there. I see a woman finally realizing her self-worth, who is at peace with herself, and is preparing herself to be a leader in her community.
Sandra, thank you for sharing such a thoughtful and honest reflection on your experience with AI-generated selfies. Your observation about the AI's tendency to default to stereotypical Indigenous features despite your specific prompts raises critical questions about how these systems are trained and whose perspectives are embedded in them.
ReplyDeleteThe tension you've identified between the AI's algorithmic assumptions and your lived experience as an Indigenous woman speaks to larger issues about representation and power in digital technologies. Your comment that "each new generated image seemed to lose the Indigenous aspect of my identity" is particularly compelling. It raises important questions: What does it mean when technology can only "see" Indigeneity through stereotypical markers? And what gets erased when we try to communicate our full, complex identities through systems that may not have the epistemological frameworks to understand them?
I appreciate how you ultimately chose to focus on emotional qualities ("happy, peaceful, strong") rather than continuing to negotiate with the AI's limited understanding of physical representation. Your final interpretation of seeing yourself as "a woman finally realizing her self-worth, who is at peace with herself, and is preparing herself to be a leader in her community" beautifully demonstrates how we can reclaim agency in these digital spaces, even when the technology itself presents obstacles.
This will be rich territory to explore further in your critical analysis. I encourage you to consider how Indigenous knowledge systems and perspectives might offer alternative frameworks for thinking about artificial intelligence and digital self-representation. There is growing scholarship on Indigenous approaches to AI that challenges the Western epistemologies currently dominating the field.
Looking forward to seeing where you take this analysis in your upcoming work. I think you'll find this article of interest as well and potentially useful to your next assignment: https://rdcu.be/e2JUi Lewis, J.E., Whaanga, H. & Yolgörmez, C. Abundant intelligences: placing AI within Indigenous knowledge frameworks. AI & Soc 40, 2141–2157 (2025). https://doi.org/10.1007/s00146-024-02099-4