Created by Dipika Titus, Mathew Olson, and Dror Margalit
Imagine a world where everybody can communicate in every language. There is no need to learn a second language, no miscommunications, and no misinterpretations of people's intentions. All of these would be bridged by our mutual understanding that language lives within a social and cultural context. Now, look at the computer that mediates this text to you: it is incredibly close to creating such a world.
I'm Smart In My Own Language is an AI-generated image collection presenting a world where synthetic visual imagery bridges cultural differences. It was inspired by the idea that sophisticated thoughts are challenging to communicate across different languages. During our brainstorming process for this project, we discussed with a colleague who told us, "I'm Smart In My Own Language," as she expressed her frustration from communicating regularly in her second language.
To make this collection, we asked native speakers of 7 different languages to describe the same set of "real" images in their native tongue. Then, we connected Google Translate to the AI image-generator Stable Diffusion, so people could write prompts in any language to generate images.
In other words, people communicated with an AI system which communicated with another AI system to produce an image. The result is a poster filled with images that look similar to one another but have but tell slightly different stories.
When approaching this project, we could not know how it would look. Still, looking at the outcome in the context of some of the current discourses around media technologies presents an optimistic picture of AI technologies. We chose to embrace the cultural differences that AI struggles to represent. Ultimately, we hope that this work will inspire the kind of reality where synthetic visual imagery helps us bridge communication differences—building a world where everybody can understand every language.
Zeynep Elif Ergin
Special thanks to: