Society for Cultural Anthropology
Oral Presentation Session
This paper examines the recent emergence of “attention” as a matter of concern in two apparently disparate parts of the software industry. First, a relatively new line of critique, emerging primarily from actors within the discursive world of Silicon Valley style itself as a defense of the “human” against technological incursions. The “human” imagined by this humanism—embodied in organizations like the Center for Humane Technology—is centrally defined by the capacity to pay attention; the problem with much contemporary software, according to this critique, is that it fails to respect human attentional sovereignty, interpellating users as instinctual animals rather than rational humans. While this attentional humanist critique is gaining purchase with tech executives, attention has also emerged as a peculiar object of interest within the technical world of machine learning: neural networks, loosely modeled on a loose model of the neuronal structure of the brain, are now being equipped with “attention mechanisms” that intentionally constrain their function to imitate human cognition. Where machine perception once promised unbounded scope, free from human attentional limits, now these limits are being emulated as potential enhancements to pattern recognition software. Borrowing its title from a technical research paper on attention mechanisms, this paper investigates the contemporary valences and values of “attention” in technological worlds. What does it mean that, for an influential set of critics, attention is coming to define the human just as it is being implemented in software? What understandings of attention, humans, and software does such a conjuncture depend on?