White Box,

Archival Inkjet Print, 14x21”,

$444

 

A white box describes a code that can be seen, but is so complex it is functionally equivalent to a black box. Neural networks, such as the visualization seen as background in this image (thanks google), are complex sets of algorithms that can be “trained” to sift through vast troves of data to find…well whatever you train it to find. You provide the input, say, a face, and the neural network will provide the output-- classify it, tell you what it is. Ideally, if you provide a human face, the network will identify it as a human face (see “google’s gorilla problem”). How exactly it comes to these conclusions is not known, hence the box.

 

“Likewise reduced to its basic elements, the camera’s program is as follows: first, to place its inherent capabilities into the image; second, to make use of a photographer for this purpose, except in borderline cases of total automation (for example, in the case of satellite photographs); third, to distribute the images produced in this way so that society is in a feedback relationship to the camera which makes it possible for the camera to improve progressively; fourth, to produce better and better images. In short: The camera’s program provides for the realization of its capabilities and, in the process, for the use of society as a feedback mechanism for its progressive improvement.” p.46