Gender synths, 2021

In an era where gender identity is increasingly recognized as complex and fluid, it seems paradoxical that computers can so readily categorize gender.

I found a dataset containing 90,000 gendered synthetic images and began experimenting by merging them back into singular faces. My goal was to explore how the model had "gendered" each image. These mirrored pairs represent a step in understanding this process.

These faces don't belong to real people. They are composites made from tens of thousands of Flickr photos. I think of them as remixes of photographic patterns, appealing to our desire to perceive photographs as real.

I aim to contrast these images with a more nuanced discussion of gender. Additionally, I'm interested in comparing the technology that facilitates gender specification, particularly when creating images of gendered faces versus gendered bodies.