CAN A PLEXIGLASS BOX OCCUPIED BY ROBOTICALLY MOVABLE CUBES AND GERBILS SIMULATE URBAN DYNAMICS?

 

The installation, entitled SEEK, was part of Nicholas Negroponte’s Architecture Machine Group experiments for the exhibition “Software” at the Jewish Museum in New York in 1970 curated by Jack Burnham . The Architecture Machine Group’s goal, which later became the MIT Media Lab, was an attempt to mediate abstract visualization models in the computer and material conditions in physical reality by enclosing gerbils in a physical Plexiglas container with five hundred zinc-plated cubes which were moved with a robotic arm.

The computer management system was developed to sense the environment and maintain order for gerbils that would act as both test subjects and constraints for the computerized system. SEEK tested the dialogue between two intelligent systems in a closed feedback loop of communication, moving past the execution of instruction (code) into learning and predicting through cybernetic models.

Negroponte’s goal was to establish a life sensing computerized system that would help the architect engage with the complexity of urban dynamics. According to Negroponte, the machine’s creativity was seen as inseparable from the designer’s initiative and thus the aspiration in this contained livable container was to develop an “evolutionary machine” ( Negroponte, “Toward a Theory of Architecture Machines,” 9). If computers have so far been seen as automatic operations through programming languages determined by designers, Negroponte attempted to initiate a “dialogue between two intelligent systems—the man and the machine” ( Negroponte, “Toward a Theory of Architecture Machines,” 9); he tried to move past the execution of instruction (code) and into learning and predicting through cybernetic models.

A cybernetic model would enable architects to control and communicate with a non-static entity that itself could not be programmed—the gerbils. The project was successful in exploring the possibility of an automated architectural, form-building environment, but derailed from its original goal: to create an environment that adapts to the whims of its occupants. Initially, the gerbils outwitted the machine, taking joy rides on the robotic arm, and carving pathways through the blocks without actually using them as nests (Hess, “Gerbils ex Machina,” 23). Later, they “were pressed together with ennui” and “terror”, similar to “shipwreck victims after 30 days in an open boat.”(Hess, “Gerbils ex Machina,” 23)
 

Does this imply humans, with regards to urban design, agitate the existing condition thus developing a need for a custodial mechanism to respond to human actions? Also referred to as “Blocksworld”, this cybernetic model reminds us that human intelligence is a misidentified feature not prescribed in the idealized cycle between action and response; SEEK itself, developed an unexpected, fragile and incoherent form of intelligence between its involved parties, its own closed world, which still does not foster in the least our ardent enthusiasm for smart devices.

 

KEYWORDS: Robotic Arm, Machine Learning

​

KEY FAILURES

​

PSYCHOLOGICAL EFFECTS: The robotic arm had a more obvious psychological effect on the gerbils than a physical effect on their environment.

​

OUTSMARTING MACHINE: The gerbils began to outsmart the machine by climbing on the robotic arm.

​

PUBLISHED MATERIALS: There were no published results which explained what the machine ‘learned’ from the gerbils.

​