Extinction Event
Trigger Warning: police brutality, violence against Black people
I served as a research assistant for ayodamola tanimowo okunseinde’s project Extinction Event, a series of experiments that deconstructs the linear temporal narrative that oppresses Black bodies and uses technology to reimagine Black temporalities.
Extinction Event computationally creates images of Black death at the hands of police and presents them with algorithmically-generated text based on Black novelists, philosophers, and critical race theorists. I helped gather image data and conduct qualitative research related to police violence against Black people in the United States and tested different natural language processing models (GPT2-simple, textgenrnn) and generative adversarial networks (DCGAN, StyleGAN) to create text and images.
Created in collaboration with ayodamola tanimowo okunseinde, Sohee Cho, Jeffrey Geiringer, and Nicole Lloyd
Data collection
We created a database of incidents of Black death at the hands of police that included date and location information, as well as media images and footage.
Generative Adversarial Networks
Using the DCGAN architecture, we fed the media we had collected into the model and produced hundreds of computationally generated images like the ones above (images have been blurred and darkened due to violent content).
Natural language processing
We also scraped and cleaned hundreds of texts from writers like Octavia Butler, Frantz Fanon, and James Baldwin. We fed these texts into GPT-2 in order to produce snippets of computationally generated text from black writers and theorists.