Working Title:
Bias In The Machine
Aims and Objectives:
My proposed plan of study will examine the relationship between technology and marginalised sectors of society, in particular UK based black men. My research will explore and unpack the attitudes, systems and tools which have and continue to create disparities between the experience of the black user in comparison to the experiences of his white counterpart.
Context:
Around the 1940’s Kodak developed a tool called the Shirley Card which was used to calibrate their film stock. The models featured on the Shirley Cards were originally white and this inevitably lead to their product being unable to capture dark skin effectively in many cases. This “oversight” serves to illustrate that black and brown people had been left out of the equation in favour of the default, the “Normal” in Kodak’s words. The assumption being that photography was the pastime of the white middle and upper-classes.
Fast-forward several decades and we find ourselves making similar mistakes, but with questionably more sinister motives. In 2015 Google was forced to issue a public apology when their photo app mistakenly labelled a black couple as gorillas. Four years later New Scientist magazine revealed that the Home Office knew its automated passport photo checking service would perform poorly on dark skin, but they rolled it out anyway. Two examples of systems largely designed by white men and failing to acknowledge the value of training data representative of a true cross section of the population.
Disparities like these and the attitudes that feed into them interest me and will form the backbone of my study.
Methodology:
My investigation will combine qualitative and quantitative data taking into account findings from organisations committed to narrowing the technology gap as well as those who seek to ensure fair and ethical use of such technologies. I’ll focus on the UK although I may take into consideration testimonies and findings from other geographic areas. First-hand experience with automated systems such as those employed by HM Passport Office and AI photo generators such as Leonardo AI will form part of my study. I’ll also consider observations from a sample of black men in the form of a simple survey based on lived experiences.
In summary, I intend to base my research on elements of:
· Documented evidence
· Personal evidence
· Survey group findings
Outcomes:
· Art project
A series of collages which comment on the issues faced by people of colour when interacting with AI based systems.
· Installation
An audio/visual project based on the outcome of the survey.
Work Plan:
Weeks 1 To 20 - RESEARCH
· Reading
· Documentation
Weeks 21 to 40 - DEVELOPMENT
· Formulation of survey questions
· Conduct survey
· Data analysis
· Artwork production
· Written response
Weeks 41 to 60 - EXECUTION
· Bringing together elements of the project
· Production
· Exhibit
Bibliography:
Ada Lovelace Institute
Website
Algorithmic Justice League
Website
Centre for Data Ethics and Innovation
Website
Coded Bias
Documentary / Shalini Kantayya
Considering human imagination the last piece of wilderness, do you think AI will ever be able to write a good song?
Article / Peter Ljubljana / The Red Hand Files
How Color Film was Originally Biased Toward White People
Article / Michael Zhang / PetaPixel
How well do IBM, Microsoft, and Face++ AI services guess the gender of a face?
Video / Joy Buolamwini and Timnit Gebru
To Photograph the Details of a Dark Horse in Low Light
Blog / Broomberg & Chanerin
The Politics of Images in Machine Learning Training Sets
Article / Kate Crawford & Trevor Paglen / Excavating AI
Race After Technology: Abolitionist Tools for the New Jim Code
Book / Ruha Benjamin / Polity
Scary Smart: The Future of Artificial Intelligence and How You Can Save Our World
Book / Mo Gawdat / Pan Macmillan
Study finds gender and skin-type bias in commercial artificial-intelligence systems
Article / Larry Hardesty / MIT
Tay, Microsoft's AI chatbot, gets a crash course in racism from Twitter
Article / Elle Hunt / The Guardian
Tech Talent Charter
Website
UK launched passport photo checker it knew would fail with dark skin
Article / Adam Vaughan / New Scientist
Unconscious Bias
Book / Annie Burdick / Summerdale
Weapons of Math Destruction
Book / Cathy O’Neil / Penguin
Resources:
Craiyon
AI image generator
HM Passport Office
Passport service
Leonardo AI
AI image generator
Comments