The widespread adoption of computer vision affects the lives of its users at unprecedented levels. Users are unevenly impacted and the populations already exposed to various forms of discrimination are the most vulnerable.
We address the question: How is seeing being taught to machines?
Our point of departure is John Berger’s 1972 BBC documentary series and book Ways of Seeing, which had an enormous impact on both popular and academic perspectives on visual culture. Central to our proposal is Berger’s assertion that: “The relation between what we see and what we know is never settled.” The project asks in what ways does computer vision further unsettle this relation. What does the increasingly general application of AI and machine vision algorithms in platforms and software imply for human ways of learning to see? What is the relationship between learning to see and training machines to see? The research proposes to investigate what new forms of computer visual literacy might be necessary when machines intervene in the interpretation, circulation and production of images. It brings together different approaches to visual and media literacy situated in the contexts of art education, pedagogy and computer science.
Through hands-on experimentation with dataset curation and machine training, workshop participants are given a concrete context through which they can understand the risks and benefits of AI and computer vision.
The interest is how visuality has been transformed by developments in computer vision and how particular elite forms of knowledge are legitimated to support intersectional discrimination and exclusion. We aim to inform a public understanding of these processes and to expose the limited worldview of image datasets, and to include other voices in this discussion.
Toolkit in development
Collaborators
Learning Experiments in Computer Vision and Visual Literacy is a public engagement project funded by The Alan Turing Institute, as part of a research collaboration between the Centre for the Study of the Networked Image (CSNI) at London South Bank University, UCL Institute of Education, and The Photographers' Gallery. With contributions from Geoff Cox (CSNI), Annie Davey (UCL IoE), Yasmine Boudiaf (Justice Matrix), Nicolas Malevé (Constant/Aarhus University), Janice McLaren (The Photographers’ Gallery), Yugyoung Choi (UCL IoE), Hsin-Mei Lin (UCL IoE), Siddony Kalair (Roding Valley High School), Makaila McKenzie (Heartlands High School), James Stevenson, Andrew Dewdney (CSNI), Tim Fransen (CSNI), Dean Kenning.