Towards an Adversarial Artistic Inquiry of Generative Computer Vision
Project dates (estimated):
September 2023 - Jan 2027
Name of the PhD student:
Martin Disley
Supervisors:
Professor Ewa Lugar – Edinburgh College of Art
Dr Morgan Currie – Social and Political Science
Beverly Hood – Edinburgh College of Art
Project aims:
This research project aims to develop and analyse a particular visual practice of artistic inquiry characterised by adversarial interventions with generative AI applications. Using a portfolio of new media work as a series of case studies, this project develops a theoretical contribution that offers a new perspective on the aesthetic, epistemic, evidential and translational value of art and design work that interrogates the ethical and cultural implications of generative AI.
The practice at the centre of this study is characterised by an intention to socialise the problematics identified in the humanities and social science scholarship on generative AI systems with either the public or with data scientists and engineers. It's methods, focussed on acting back on the model to evidence or demonstrate these arguments, are more closely aligned with engineering practices. This interdisciplinary practice combines direct material investigation, by importing methods from computer security and the practices of red teaming, and the affective and experiential potential of aesthetic practice to produce cultural artefacts that evidence the problematics of these systems.
Disciplines and subfields engaged:
Research through Design
Technology Ethics
Media Art
Research Themes:
Adversarial Computing
Red Teaming
Computational Attacks
Ablation Studies
Research through Design
Provotyping
Boundary Objects
Translation
Related outputs:
Disley, M., & Khan, M. (2024). Not I: The Voice, Identity and the Epistemic Mirage of Machine Learning. xCoAx 2024: Proceedings of the Twelfth Conference on Computation, Communication, Aesthetics & X., Fabrica, Treviso, Italy.