We hope for these conceptual tools to inform the fields AI ethics & Responsible Innovation, as a way to study how AI ethics principles are translated into the material processes of knowledge production, highlighting sites of epistemic injustice and potential for intervention 4/4

Building on David Beer, we introduce the concept of Representation Coils, looping effects with downstream implications where these assumptions concretise labour dynamics. RC help us understand how and where classifications of skill and intelligence are formed & maintained 3/4

Specifically, we look at the practical contexts of this classification and division of labour—the points of contact between AI practitioners and data workers—and illustrate how practitioners’ representations of data work can propagate the labour and power structures that support AI development. 2/4

We bring findings from our empirical study in dialogue with critical scholarship on AI and data work to explore the ways specific epistemic legacies and supply chains of labour shape AI practice & governance 1/4

Ooh this is on my reading list!

I’m very sorry for your loss, Shannon 💜