5 - Emptied Faces: In Search of an Algorithmic Punctum
Published online by Cambridge University Press: 18 November 2022
Summary
INTRODUCTION
File 28_0_3_20170119194439188.jpg presented a familiar face in a new context. Reading the nomenclature of the file, I was to encounter the face of a twenty-eight-year-old Asian male as part of the UTKFace dataset. Yet, upon looking at the image, I encountered a familiar face – namely that of the infamous Afghan girl Sharbat Gula who graced the cover of National Geographic, inspired support for Afghan refugees and became the hallmark of the difficulties of migration, war and displacement. Reduced to a data-point, Steven McCurry's photograph of Gula and Gula's face itself were to be emptied of any prior cultural meaning and rearticulated in a study about the ability of algorithms to predict ageing patterns. Gula's face was now to serve as nothing more than a set of facial landmarks.
In thinking about how the face becomes reconfigured in the context of algorithmic culture, this chapter engages with the politics of emptying photographed faces of meaning and their rearticulation into data-driven predictive landscapes. More specifically, it explores how photographic portraits have been scraped from the World Wide Web and further reduced and classified by what Deleuze and Guattari call ‘faciality machines’. In the aftermath of this classification, the images themselves emerge as what Mitra Azar has called the ‘algorithmic facial image’ and faces are further reduced into facial parts or facial attributes. I trace the life of the face as raw data in the context of the UTKFace dataset by deploying the methodology of ‘archaeology of data’ articulated by Kate Crawford and Trevor Paglen in their project ‘Excavating AI’. Further, I argue that the decomposition of the face in the context of big data and machine learning algorithms has led to two distinct rearticulations: one linked to predictive algorithms and the other linked to the generation of deepfake portraits. Last, but not least, I argue that ‘faciality machines’ might be resisted by thinking about the face in relation to what Roland Barthes calls its unanalysable air – air that could prick us, and therefore has the potential to act as a punctum, by prompting an affective relationship to the images, faces and people that stand before us.
- Type
- Chapter
- Information
- Faces on ScreenNew Approaches, pp. 75 - 90Publisher: Edinburgh University PressPrint publication year: 2022