My interest in machine learning began during my undergraduate education through conversations with my partner who was studying Computer Science. The concept of a machine mimicking human behavior fascinated me, and I began to consider the ways in which I might incorporate machine learning into my artistic practice. In the last three years or so, my work began to develop significantly in response to my engagement with neural networks.
As a painter working with AI, I span a variety of mediums and methods. My abstract paintings tend toward the organic, characterized by expressive color and baroque compositions and featuring molecular, fluid, and constellation-like forms that develop from repetition and visual association. While my influences constantly evolve, I continue to be inspired by biology and botany and work within a balance between intuition and structure.
Rebecca Mott, Mutation 108, 2020.
Leading up to my first semester in SAIC’s Low Residency MFA Program, I began to experiment with a variety of tools and discovered Deep Dream, from which my Mutations paintings originated. Through this project, I began to tease out ideas about machine learning and the effects that it might have on contemporary painting and society's view of it. Using photographs of my existing paintings and Deep Dream’s “Deep Style” system, I was able to merge two of my original paintings into one. The Convolutional Neural Network identified the colors and patterns of one image and then transplanted them onto the compositional framework of the other image, extrapolating details along the way. I loosely transcribed the digital images that resulted from this process into paint on canvas, allowing my limitations of observation and replication to recreate the image.
After completing a series of ten paintings, I returned to a query from months before about identity and authorship. In 2020 my partner and I began planning and construction of what we refer to as the Doppelgänger Project. Our long-term objective is to craft a biologically distinct double of my artistic identity in the form of a GAN, or General Adversarial Network.
Rebecca Mott, Training Set A Paintings 1-3, 2021.
At the foundation of this project is the Training Set A series, which consists of spontaneous and impromptu paintings intended to communicate and codify my raw, creative vision while maintaining maximum flexibility and authenticity. These images are fed to the General Adversarial Network, which is designed to replicate my unique style and simulate individualized artistic consciousness. I completed the Training Set A series in 2021, ending with close to one hundred and fifty new paintings, and I have revitalized a more organic painting practice since then. All paintings that I create from here on will be added to the training set for the Doppelgänger GAN.
We have continued to work on the Doppelgänger Project with the hope that it will eventually be capable of producing images that seamlessly identify and reflect my artistic tendencies. Because of limitations in our data and current resources, we have not yet achieved our desired output. This is an ongoing process that we will continue to work on for the foreseeable future while engaging with other tools and opportunities in the meantime.
One of the essential purposes of the Doppelgänger Project is that it would be personally constructive and challenging, a system that is able to inform and expand my artistic practice as I build upon the precedent of those who have taken up the mantle of new technologies for the benefit of all artists. Entangled within that is my desire to interrogate theories of gendered labor and social standards regarding the authenticity and value of art that is not human-made.