FACES
F4CES
F4c3S
PH4c32
Every face embodies a paradox: A presence that is absent; an absence that steps forward and binds. Faces are everywhere and nowhere, near and far, familiar and completely strange: force fields of recognition – and masks of death. Every serious physiognomy (whether in philosophy or art) thus begins with a critique of the face, with a careful unveiling of the cryptic – never-seen – sign on one’s own body.
Thomas Macho, „translated from ‚Gesichtsverluste’" 1996
Traditionally, the face was an expression of individual personality and a mirror of mental and emotional states. Today, the face is reduced to mere technical data through digital technologies like facial recognition software. This emptying of the face is evident in the digital age, which treat the face as an "interface" or mere surface with no deeper meaning or individuality. Art historian Hans Belting describes the face in digital society as a "surface" detached from its physicality, where faces come and go like mask-like representations.
This transformation of the image of the face is driven by three key factors: First, digitalization, which changes the truthfulness of mass media images and produces algorithmically manipulated image products. Second, new facial recognition technologies that enable a control-based representation of reality. Third, the surveillance measures following September 11th, which led to the political and economic instrumentalization of facial and body surfaces.
In the course of digitalization, the face becomes increasingly supra-individual and mask-like, losing its authentic core. While Emmanuel Lévinas saw face-to-face dialogue as the foundation for an ethical and democratic society, this direct encounter is now replaced by superficial interaction with avatars and facial surfaces. The mass reproduction of faces—such as through photographic passport photos—has also reinterpreted the individuality of the face in favor of standardization and state control. Photographed faces became an instrument of police and registration archives, used to standardize and bureaucratically record faces.
Modern surveillance, particularly in the context of "liquid surveillance," operates through algorithmic, often invisible systems that exploit data users unknowingly provide to platforms like Google or Facebook. These platforms are developing increasingly powerful facial recognition software capable of recognizing and identifying faces in massive image databases. These data are used by intelligence agencies and private companies to locate individuals based on their facial features and track movement patterns in public and private spaces.
When we look into the inner workings of machine vision systems, we encounter a multitude of abstractions that seem completely foreign to human perception. The world of machines consists less of representations and more of activations and operations. These systems are based more on active, performative relationships than on classical representations. However, this does not mean that there are no formal foundations for their functionality.
All machine vision systems generate mathematical abstractions from the images they analyze. These abstractions are determined by the type of metadata the algorithm is attempting to interpret. In facial recognition, for example, different techniques are used depending on the application, efficiency requirements, and available training data. An older method analyzes a face by filtering out the features that match other faces. What remains is a unique "fingerprint" or "archetype" of the face. To recognize a specific person, the algorithm then searches for the fingerprint of that face.
"Seeing machines" is a broad definition of photography. It is meant to encompass the countless ways in which not only humans use technology to "see" the world but also how machines see the world for other machines.
A „script“ behind it, as the fundamental and obvious function of an imaging system, its "mode" of seeing, and the immediate relationships (e.g., between seer and seen) that it creates, as well as the evident ways in which a "seeing machine" shapes the world. Simply put, a script encompasses all the things a particular "seeing machine" "wants to do," how it "wants to see" the world, and how it does what it is designed to do. A machine's script strongly limits certain activities and relationships while excluding others. Here, there is a dual question: The "how" of a machine's seeing is inextricably linked to the effects it produces.
by Arne Flemming and Lucas Knäuper
