Doe no. 4

video: 1504  x  2152 / 13s loop physical: 18” x 22” x 1.5” framed fine art print

Doe no. 4 is a portrait from my Doe series, an algorithm’s representation of humans. This Jane Doe portrait is an anonymous, amorphous machine hallucination created by a generative machine learning model.

I trained the model exclusively on artwork by women artists. Art history has been dominated by male authorship, and machine learning models perpetuate the biases of the data used to train them. This ethical problem sits at the core of machine learning: when we train models using historical data, how do we counteract the biases embedded in the data? Pushing back against the one-sided current of art history, we can create the possibility for new visions.

20% of proceeds will be donated to Girls Who Code to support their mission to close the gender gap in technology.