As of December 2022, over 1,080,000 people in the United States have died of complications due to the SARS‑CoV‑2 virus. Amongst the rampant misinformation and toxic political environment that has characterized the fight against the COVID-19 virus, there exists little room for honest self-reflection. How does one begin to mourn such a loss? How do we, in our collective consciousness, grieve that which is nearly impossible to comprehend?
In my work, Loss Beyond Comprehension, I utilize an emerging artificial intelligence technology called a Generative Adversarial Network (GAN) to create hyperrealistic images of human faces from photographs of USA COVID-19 victims. By compiling a dataset of thousands of images of real COVID-19 victims, I was able to train a GAN to synthesize new images of people that do not exist. In this way, each face is a synecdochical representation of the thousands of lives lost.
These portraits are created by the synthesis of no longer existing people into people who have never existed. They are anonymous figures residing in a latent space between life and death, real and fake. Just as many Americans tragically die without any family or friends to carry their memory, so too do these portraits possess a sort of mirrored anonymity. They are both everyone and no one.
With the rise of machine learning and ever advancing AI technology, the line between what is real and fake is continuously being blurred. We have seen how Deepfakes and similar synthetic imaging technologies have been used to intentionally deceive constituencies, sow political chaos, and influence the outcome of elections. And while there is a pressing need to grapple with these ethical issues, I believe there is an equal need for finding ways to use this technology to benefit the greater good. This work represents my contribution to this ongoing effort.