I'm delighted to present 'Florilegium', a new four part multi-layered installation and public program at the University of North Texas. The CVAD gallery exhibition includes 'Latent Garden', a series of 36 machine learning paintings and 'Active Line', a video installation featuring music by Shara Nova. It is open from Aug. 24–Dec. 10, 2021.
There will also be two performances of 'Infinite Movement', a newly commissioned work with music by Shara Nova, a live outdoor performance on Oct. 9, 2021 and an indoor presentation on November 2, 2021.
'Shadow Garden', an Art in Public Places permanent commission for the University of North Texas will be installed between November and December 2021.
The project will culminate with a digital publication documenting the exhibition, musical performances, sculpture, class, programming, and the collaborative efforts of all those involved, as well as a recording of 'Infinite Movement'.
Originally, a florilegium meant a collection of canonical texts, or 'flowers of wisdom'. Over time, the word became more commonly associated with collections of lavish, often hand painted, botanical images. This exhibition contains something of both those legacies. The once ordered flower garden of canonical knowledge has become an overgrown meadow of wild information. The COVID-19 pandemic, during which this project evolved, only accentuated the omnipresence of the internet, an information space characterized by unstable connectivity, conspiracy theories and incompatible public and private priorities, all illustrating the many ways that stable informational supply chains are collapsing. This wild meadow, choking on overgrowth, is surrounded by an unknowable forest of data, filled with predatory algorithms and the emerging large scale computational entities known as 'Artificial Intelligence'.
A closely related strand of my own research began a few years ago at MIT with the incredible team at Center for Art, Science and Technology, where Dr Sarah Schwettman introduced me to 'generative adversarial networks' or GANs, a form of machine learning frameworks that emerged over the last decade. Converting very large archives of images into data, the GANs compare variations based on similarities and differences, creating a mathematically vast 'latent space'. What the GAN searches for in this space depends on how the original training set of images is defined. Often the GANs are looking for something like the common features that define a widely known image - like a human face - so they can generate 'new' versions of the old data. When they cannot generate a match, they generate a recognizable and visually distinct form of machinic production noise, a visual 'glitch' relatable to the print production 'errors' found in Andy Warhol and Wade Guyton and recently explored by Pierre Huyghe in his 'Umwelt' project.
For me, it is significant to note the GANs do not simply morph between a library of existing forms but generate new material allowing their innate ability to generate formal variation to be stimulated into forming seeds for exotic new varietals. This prompted the consideration of a 'sketch space', an unfinished threshold zone between the final finished image and the purely informational terms of latent space. Inspired by an essay I was writing for Kelly Chorpening's, 'Companion to Contemporary Drawing', and with Dr Schwettman's help, I created a series of mini-GANs using training sets of historical sketches, combined with the animation of the sculpture design for UNT. While the imagery for the paintings developed, another collaborator I encountered at MIT, the artist and inventor Ben Tritt, had developed a sophisticated painting machine for his company Artmatr (a story that fully deserves its own blog post another time). Without this technology and the generous assistance of Artmatr, these paintings could not exist in the liminal state between theoretical information and real body, between printer and painter that seems essential. This doubled process, moving from neural network to canvas, combining past and present technologies, generated a large series of unique and surreal artworks, a 'latent garden', filled with informational branches, mycelial networks, and data blisters, encoding the wild contemporary garden of proliferating information while visually and thematically echoing the current paradox of stasis and change in which we find ourselves.
A third part of this collaborative project grew later, in the form of a new choral work, 'Infinite Movement', composed by UNT alumna Shara Nova, (B.A., Performance - Voice) for the UNT university singers, and directed by Associate Professor Kristina MacMullen, D.M.A. This extraordinary and gorgeous work explores a range of choral possibilities, reflecting in sound the learning processes of the GAN and the students, finding connections between disparate sonic strands and generating and sustaining what might exist in the spaces between them. My lyrics for 'Infinite Movement', combine GAN generated texts along with edited fragments from John Milton's Paradise Lost and Paul Klee's pedagogical sketchbook, (the original inspiration for the UNT sculpture. 'Active Line'), and an accompanying animated film in the gallery includes fragments of the 'Infinite Movement' sound work as well as animated imagery from the GANs.
In this spirit, I also collaborated on and co-created the fourth element, an artwork made with students from the Topics in Contemporary Art Practice class, co-taught with Stefanie Dlugosz-Acton, (director, CVAD Galleries), in Spring 2020. As part of the exhibition, two videos that were created for the course — a co-created GAN artwork from the class and 'An Artist Lecture' by Sean Lopez, (a darkly humorous look at my own teaching practice that made me laugh until I cried) — can be seen projected on the northeast side of CVAD at the corner of Mulberry and Welch streets, after dusk throughout the exhibition.
Still from the 'Active Line' animation, showing GAN interpretations of the UNT sculpture and the 'Latent Garden' source imagery. Matthew Ritchie Studio, 2021