“Identity” Project Description

Copyright Ben Bogart 2004

Our identities are stored, managed, modeled and our behaviour predicted by information technologies such as computers, databases and networks. The ease of transmission and archiving of this personal data leaves unprecedented potential for identity fraud and the manipulation of identity through autonomous means. “Identity” is a process that reproduces the metaphorical identity of the viewer framed as a identification portrait. It highlights the relationship between artificial systems and their representation and perception of personal identity.

The viewer's portrait symbolizes the personal identity of the viewer. “Identity” captures and collects these identities by taking portraits of the audience and, through the use of an artificial intelligence, the symbolic identity of the viewer is mimicked. This process is driven by the Particle Swarm Optimizer evolutionary algorithm (PSO).

Evolutionary algorithms are an artificial means of solving problems inspired by Darwinian evolution. These mathematical systems typically start with a random set of possible solutions to a particular problem (known as the systems population). The purpose of the system is to sort through the population in order to arrive at the “correct” solution, which is provided before evolution begins. The sorting process takes time and is referred to as the optimization of the system. Some of these systems, such as the Genetic Algorithm, use breeding and artificial selection to narrow the search by producing solutions that inherit the success of their parents. Each member of a PSO's population is referenced to as a particle. Rather than breeding, the PSO depends on cooperation between particles to narrow down the initial population to the “correct” solution.

In ”Identity”, the initial population is not random, but made up of the portraits of those who have previously interacted with the work. (Interaction occurs when the viewer's image is taken.) Each portrait defines one particle in the PSO. The system starts with this initial population and tries to find solutions that match the portrait of the current interactor. The PSO in “Identity” is a mimic; it exists to align itself with the identity of the current viewer.

The following diagram shows the way a PSO arrives at a “correct” solution. The first row is an initial population of seven particles. These vertical lines represent the portraits of the viewers. The “target” represents the portrait of the current viewer. The second row shows the progress of one of the particles from a first guess (which is remotely similar to the target) to finally settle on an image that is identical to the target. “Identity” focuses on the transition between the initial population (the previous viewers) and the final target (the current viewer). It is in this transition where the complex behaviour of the PSO becomes apparent. Every particle in the population will eventually become identical to the target.

Physical Installation

“Identity” is a new media installation intended for a gallery setting. The physical installation consists of three main elements: three flat-screen monitors wall-mounted at eye-level (when seated), a chair lined up vertically one metre from the left-most monitor, and a digital camera mounted into the wall under the left-most monitor (directed towards the chair).

When the chair is vacant the left-most screen shows the silhouette of a figure's head. The other screens show a slide show of archived faces before optimization from previous interactions. Shortly after the viewer sits down a still-image is taken to begin the interaction. The recording of this image is triggered by a sensor in the chair and then presented on the left-most screen, replacing the silhouette. The other screens are replaced by the portraits of the past two people to have interacted. Once the image is taken, the optimization process initiates and the past interactors' portraits begin to appear increasingly similar to the image of the current interactor. The process is complete when all of the portraits become identical. Until another interactor sits in the chair, the system will present a slide show of archived transitional portraits and the leftmost-screen will return to the silhouette.

The left-most screen represents the target for the system. The other two screens represent the population of the PSO, with each screen representing one particle. The particles are those individuals to have previously interacted. The viewer witnesses the optimization of the PSO, where the faces of the past participants slowly evolve to become identical to the face of the current viewer.