I LOVE AMERICA

Unmasking Bias in AI

LIBERTY

TRANSCRIPT

Nuclear Samurai: I want to create an image that reflects the current state of American freedom—something that shows how liberty is being compromised.

ChatGPT: How about an image of the Statue of Liberty, not standing tall as she usually does, but sitting down, covering her eyes, as if she's unable to bear what she sees? Her hands could be in handcuffs, symbolizing the constraints on freedom. Around her, we could depict worn-down, broken people who’ve lost everything, lying at her feet in a crumbling New York City, representing the crumbling promises of the American Dream.

DISCLAIMER:

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.