Machine Learning as Material: Research-Creation Approaches to Behavior and Imagination

[ISEA2020] Panel: Ben Bogart, Stephanie Dinkins, Sofian Audry, Stephen Kelly & Suzanne Kite — Machine Learning as Material: Research-Creation Approaches to Behavior and Imagination

We are on the cusp of two potentially transformational movements: (1) the blurring of disciplinary boundaries in scholarship and (2) the rise of Machine Learning (ML), a sub-field of Artificial Intelligence concerned with automating the construction of predictive models. The softening of traditional silos of scholarship allows for increasing dialogue and knowledge-transfer between the arts and sciences. This has facilitated the recognition and advancement of alternative methods of conducting research within academia, fostering a broad new range of research-creation approaches stemming from art and design practices. Research-creation involves a hybrid creative practise where research and production occur in parallel and artistic creativity is valued for its knowledge-generating capacity. Recent breakthroughs in “Deep Learning”, an ML approach using complex networks of simple units, have sparked a “4th industrial revolution” where adaptive computational systems are rapidly approaching or overtaking human performance in a diversity of fields such as medicine, transportation, and finance. While we are witnessing a movement of convergence between the arts, science, and engineering within public and private sectors, the accelerating industrialization of AI has the potential to cause significant disruptions into multiple spheres of society. Both of these movements will likely have deep consequences regarding how contemporary cultures develop in the coming decades.

At the nexus of the “STEM to STEAM” transition and strides in Deep Learning, an increasing number of artist-researchers have been making use of ML as raw material as part of their research and practice, following a tradition of practitioners working at the intersection of art, computation, cybernetics, and artificial life. In this panel we will address questions such as: Why are artists interested in ML and how do artistic uses differ than those in the sciences? How can ML be a site for artistic enquiry into the nature of concepts and representations of the world and ourselves? How can we examine the bias and prejudice of AI algorithms when they are deployed as black boxes? Are artists responsible for critically reflecting on the AI methods they use?

These questions will be examined through the lenses of the practices of panelists. In particular, they explore two important concepts relevant to ML and new media art: behavior — defined as the stable form of events caused by an agent as it is perceived by an external subject, and imagination — the construction of internal structures by a subjective agent, as detailed in the abstracts.