Zizi – queering data and explaining AI

Zizi - queering data and explaining AI
Zizi – queering data and explaining AI

Jake Elwes is an internationally acclaimed artist working with machine learning. This project sees him collaborating with the Experiential AI research theme at Edinburgh Futures Institute to develop Zizi, his most ambitious work to date.

Image from Jake Elwes' Zizi project
An image from Elwes, 2019, Zizi

Developed as part of the Experiential AI research theme at the Centre for Near Futures, the research partnership between Elwes and the Centre explores how artistic works can make AI systems more transparent so audiences can understand and critique their reasoning.

 

Research theme

Experiential AI is a new research theme and cluster originating in Edinburgh. The theme investigates the entanglements between humans and machines, and explores how experiential methods can reveal insights on future scenarios for ethical and responsible AI.

 

.

Summary of the project

Zizi is an art project exploring machine learning and sexual identity. It is also a research study addressing how artistic works can make AI systems more transparent, so audiences can understand and critique their reasoning. The hypothesis is that art can create new intelligibility for AI.

The first presentation of Zizi (see above) was commissioned as a part of the Experiential AI project and research theme. At this stage Elwes was looking at bias and representivity in training datasets and investigating ideas of queering the data. This version of the work was presented as a multi-screen video piece of faces of drag artists in constant transition, flowing between identities to expose what the network had learnt from injecting a generative neural network with 1000 images of drag performers.

Now, in a new phase of the project, Elwes will collaborate with researchers in Edinburgh to explore how Zizi can communicate the operation of machine learning. Working with Battersea Art Centre partners from the cultural sector, he will realise a theatrical show, a set of video pieces, and possibly also a TV documentary about the making of Zizi.

Through this practice, the research study will assess how an audience can understand what is happening in the work, and evaluate strategies to convey the character and operation of AI.

Indicative research questions

  1. How can experiential methods be used to explore future scenarios for ethical and responsible AI?
  2. How can an artwork develop literacy in audiences so they can understand and critique the reasoning of an AI system?
  3. How can an audience understand what is happening when they experience an artwork created using machine learning?
  4. How can art create edge cases (limits, extremes, outliers, provocations) and what can we learn from them?
  5. Can creative practice help engineers spot gaps in the current concepts, tools and approaches?
  6. How does the artist create new dimensions in their artistic practice through their encounters with AI research?
  7. How can this project enable other artists to discover new creative ideas and possibilities for their own practice?
  8. How could a performance script for Zizi translate and build on knowledge from AI science, engineering and ethics, without undermining coherency or rigour?
  9. What happens differently when this is a performance? What difference does a performance make? What can happen when a performer is working with an AI on stage?
  10. Why do this? What is the motivation for turning ideas into creative works?

 

Biographies

Jake Elwes

Jake Elwes (UK), born 1993, is currently based in London. His recent works have looked at Artificial Intelligence, investigating the technology, philosophy and ethics behind it. Jake graduated with a BA in Fine Art from the Slade School of Fine Art (UCL), London in 2017, having also attended SAIC, Chicago (Erasmus), in 2016 and Central St Martins (foundation) in 2012. Jake has exhibited in the UK and internationally, including:

  • ZKM, Karlsruhe, Germany
  • Today Art Museum, Beijing, China; CyFest, Venice
  • Edinburgh Futures Institute, UK
  • Zabludowicz Collection, London
  • Frankfurter Kunstverein, Germany
  • Bloomberg New Contemporaries 2017, Newcastle & London, UK
  • Ars Electronica 2017, Linz, Austria; Victoria and Albert Museum, London; City Loop, Barcelona
  • Nature Morte, Delhi, India
  • Centre for the Future of Intelligence (CFI), Cambridge, UK

https://www.jakeelwes.com

Image from Jake Elwes CUSP project
An image from Elwes, 2019, CUSP, taken on location in the Essex Marshes.

CUSP (2019, above) augments and disrupts the natural ecology of a location, as flocks of native birds enter a visual dialogue with artificial ones generated by a neural network (GAN) and projected onto portable perspex screens.

Experiential AI at Edinburgh Futures Institute

The Centre for Near Futures is a new research centre within Edinburgh Futures Institute and Edinburgh College of Art at the University of Edinburgh. The Centre presently leads two projects and research themes, Experiential AI, on the art, ethics and science of AI, and Citizens’ Observatories, on empowering citizens to participate in climate action. Recent projects include the GROW Observatory, the world’s first continental scale Citizens’ Observatory, and the world premiere of Zizi, a first edition for public displays presented in Edinburgh during August 2019.

Credits

The Experiential AI Pilots 2020 are a project of the Edinburgh Futures Institute and Edinburgh College of Art at the University of Edinburgh. They are developed in partnership with the Bayes Centre, Informatics, Edinburgh College of Art and Heriot Watt Robotarium. They are supported by Creative Scotland and the Data-Driven Innovation programme of the South East Scotland City and Region Deal.

Zizi is a research-led art project by Jake Elwes. The project is co-commissioned and produced by Edinburgh Futures Institute and Battersea Arts Centre. The research process is led by Edinburgh Futures Institute.

.

Share this post: