DRN WIP Workshop: Machine Learning Artworks as Systems-of-Systems

This is an event for DRN members only. You can find out more about the network here.

The DRN’s WIP Workshop series is an opportunity for members of the network to present papers on their research relating to British art histories. We are keen to encourage collaboration within our research community and hope that these WIP workshops will help researchers develop their work whilst simultaneously making all our members productively aware of new issues, ideas, directions and methodologies developing within the field of British art history.

Machine learning, the now predominant form of AI, performs an innumerable array of tasks in the world (Burrell, 2016:1). Consequential decisions are increasingly delegated to this technology (Bjerring and Busch, 2021). Seemingly inconsequential decisions are too, such as content exposure via recommendation engines (Manovich, 2018:2) which turn out to have critical implications. Machine learning is deployed in the ‘real world’ and trained there, while it still poses risks (Bunz, 2019:2). It has become politically urgent. Perhaps this urgency is why it has attracted the critical attention of artists, curators, theorists and institutions. Indeed if we are concerned about the risks associated with testing new technologies ‘in the field’ (Bunz, 2019) the gallery seems like a convincing intermediary in which to experiment with new tools. This talk with consider the impact the uptake of machine learning by artists, by taking as case studies projects which have recently undergone conceptual incubation as part of the Creative AI Lab at Serpentine

Galleries, particularly ISCRI, a collaboration between 0rphan Drift and ETIC Lab. Alasdair will outline the framework he has been developing to analyse AI artworks, drawing on theories of systems aesthetics (Burnham, 1968) and distributed authorship (Ascott, 2005:283) with the aim of making visible the obscured back-end collaboration of such works, contingent on the cooperation of not just artists but computer engineers, data scientists specialist producers and curators, as well as machines. Alasdair will argue that though we can understand these artworks are systems, we must acknowledge that they are open systems which interface with and causally implicate their host institutions, their audiences, and the wider world in which they are embedded. We only need to turn to the machine learning-driven Triple-Chaser (2019) – presented by Forensic Architecture at the 2019 Whitney Biennial – to acknowledge what is at stake when machine learning technologies are (re)deployed in an art institutional context (Cogley, 2019). Such works begin to offer an alternative model for how AI is developed and employed. This feedback, via art practice, into the system of both arts institutions and AI R&D demonstrates the potential for AI art to act as a moment of disruption: what systems-theorist Donella Meadows calls a leverage point (Meadows, 1999).