A brief background to my research about autonomy, AI and ethics.
My ongoing doctoral project, The Ethical Apparatus: The Imagination of Ethics, Autonomous Vehicles and their Human Managers (working title) unpacks the imaginaries, narratives and material realities in the construction of machine ‘autonomy’: its relationship to automation, implications for human agency, and human- machine relationships. Taking the case of autonomous driving, my research examines how ‘ethics’ is being both invoked, and also simultaneously shaped, as a measure of machine autonomy, which itself exists only as a cultural imaginary. I do this by showing the development of ‘the ethical apparatus’, a complex socio-technical infrastructural arrangement for the production of ethics as, variously, computational, designed-in, bureaucratic, regulatory or legible. The ethical apparatus is more than a Foucauldian dispositif; drawing from feminist technoscience and Actor Network Theory, I show how the apparatus is constructed and works to make the machine appear ‘ethical’, and humans complicit in the maintenance of that ‘ethical-ness’.
Situated in a Media and Cultural Studies tradition, my work is shaped by Science and Technology Studies (STS), critical algorithm and big data studies, infrastructure studies, Human-Computer Interaction (HCI) and cultural histories of automobility. In drawing on these theoretical frames, I argue for complicating machine autonomy in terms of continuities between human and non-human, rather than the Cartesian separation of body/brain and mind/matter. Thus following Suchman and Weber1, I argue that ‘autonomy’ is not an inherent attribute, but is “enacted within, rather than separable from, particular human-machine configurations” (p2). The absence of continuity as an aspect of human-machine relations forces early HCI paradigms such as the ‘human in the loop’, which persist as iterations on boundaries between human and non-human. If we are to imagine new ethico-political possibilities, I argue that we need to re-conceptualise those places where human and machine are more deeply entwined and develop new kinds of language to describe these.
My methodology is inspired by Annette Markham’s discussion of bricolage2 and my methods include document analysis (patents and academic papers outlining algorithms for moral decision-making, popular culture like XKCD comics and TED Talks, government policy reports, toolkits and workshops for ethics, standards-setting documents, and corporate communications); and traditional semi-structured interviews and site visits with lawyers, philosophers, computer scientists, automotive engineers, policy makers and technologists in Germany and in the United States. Alongside these, I report on my own rather unexpected ethnographic engagement as a consultant to a German company negotiating their own concerns about AI and ethics.