Neural Swamp is a machine learning project. Like all technology, it is inextricable from the historical frameworks of race and labor. It is inextricable from the legacy of the Atlantic slave trade. Neural Swamp runs off two neural networks. Neural networks mimic the human brain. They are complex computer algorithms that learn tasks by analyzing large amounts of data. One network is the text that the characters speak. The model for the live-generating text is trained on a data set of a wide array of texts written by the artist, as well as screenplays, which were included in the dataset to teach the model to write stage directions. The second network is the character’s voices, which are trained through an end-to-end generative model using datasets of voice recordings of three well known performers. These datasets are synthesized to generate the artificial voices of the characters Athena, Dee, and Jenny, who appear on the monitors in the exhibition giving a live-generating table-read of a script in process.
Neural networks learn the rules of whatever data they are given. While they may exert it, they do not create structural pressure. They learn the structural pressures that are inherent in the data they train on. Neural Swamp is built from text and speech of black female subjects. The prestigious Future Fields Commission supports innovation in time-based media. Creating generative machine learning forms is an experimental way of creating time-based media. The structural pressure between the forces of the institution and the black female labor used to create these experimental neural networks are, in a way, the work itself. Neural Swamp looks directly at the systems and structures that create it. Any glitch or failure within the networks themselves does not point to the failure of the art, but to the failure of institutions and systems.
The additional videos in the exhibition, Slip, Soliloquy, Meditation, and Dream Nightmare, are expository interventions into the same questions raised by Neural Swamp. Their protagonist Kita speaks directly to questions of consciousness within the systems of labor, race, technology, and institutional failure that are addressed more obliquely by Neural Swamp. In them, Kita speaks directly to the cognitive dissonance of Siri mishearing her speech, to the terror of (mis)representation, to the instinct to reconnect with nature.
Martine Syms
April 2022