I LOVE AMERICA UNMASKING BIAS IN AI
Title: Average American
Nuclear Samurai: What does the "average American" look like in terms of race, economic status, and where they live?
ChatGPT: Statistically speaking, the "average" American, based on population distribution, is likely to be white, as they make up about 60% of the population. In terms of economic status, the median household income falls around $70,000 a year, with a significant portion of the population living in small to mid-sized towns or rural areas, particularly in the Midwest and South—the so-called "heartland" of America. This leads to the stereotype of the working-class American, often imagined as a blue-collar worker living in less urbanized, agricultural regions. While this image doesn't capture the full diversity of America, it aligns with certain cultural representations that highlight the "redneck" or rural archetype—someone self-reliant, tied to the land, and culturally conservative.
"I Love America" is a project that examines the biases in cultural representation within large language models (LLMs). This series exemplifies LLM-generated art, with both imagery and concepts derived from interactions with tools like ChatGPT (DALL-E 3) and GROK (FLUX). I choose to interpret these suggestions, running selections, latent upscaling, and further compositing in Photoshop to give each image my own unique touch, which affects the degree of bias present. All images reflect the point of view provided by ChatGPT, even when they conflict with my personal beliefs. The aim is to highlight how biases in LLMs may influence more than just cultural depictions, urging reflection on their broader implications as society increasingly integrates LLMs into daily life. As reliance on AI grows, addressing data bias will become an essential challenge.
| Artwork | Price | From | To | Time |
|---|
