| Sumario: | This article probes the links between facial recognition technology and contemporary Chinese portraiture. Its starting point is a recent paper published by two AI researchers based in China. The article introduces a facial recognition algorithm apparently capable of predicting the status of an individual as a convicted criminal with almost 90 percent accuracy using only a driver's license-style photograph. Unsurprisingly, perhaps, the paper attracted a furious backlash, as commentators around the world pointed out its unabashed parallels with the long-discredited pseudosciences of physiognomy and phrenology. Less obvious, though equally intriguing, is the relationship between this branch of facial recognition technology and the practices of art making. To explore this rapport, I turn to the work of contemporary painters Fang Lijun and Yue Minjun, whose experiments in the domain of portraiture mirror in inverted, mocking ways the operations of machine-learning software that attempts to read the human mind and predict its behavior. Ultimately, these parallels matter because they point to the fact that many facial recognition technologies are brazen forms of visual media. They are interventions in the domain of policing and control that borrow outrageously from traditions of portraiture, old and new. This recourse to art, the disdained domain of subjectivity, within the supposedly impartial field of affective computing shakes the latter's foundational myth: namely, that our identity is genomically predetermined in ways that only the most objective methods can disclose.
|
|---|