Overloaded.supply – Superfluous (object) futures

Created by Ignacio Pérez (ECAL MID), Overloaded.supply is a critical tool that questions our models of manufacturing and consumption, exploiting the control of creation algorithms in order to interrogate current patterns of design, production and legislation.

At some point in 2020, we reached a critical state where the mass of man-made artifacts exceeded the biomass on our planet1. In this project the patent drawing is addressed as one of the key elements of this circle: intellectual property perpetuates a model where the surplus of artificial objects is the norm and there is a legal system in place that protects this process. ML algorithms have been democratized in the last years so the question becomes directed to how we can apply them to this criticism on overproduction; a machine intelligence system can generate objects both in speed and volume surpassing human capabilities (not necessarily quality, usability or feasibility).

Ignacio Pérez
[1] Elhacham, E., Ben-Uri, L., Grozovski, J. et al. Global human-made mass exceeds all living biomass. Nature 588, 442–444 (2020). https://doi.org/10.1038/s41586-020-3010-5

The system is comprised of three main elements: a patent drawing generation system, running on the StyleGAN2 architecture which is controlled by the second element: a physical object that allows us to customize certain aspects of the current object: its main shape, complexity or organic features. A fine tuning based that also touches the theme of making the latent space —a highly abstract concept— more approachable and encapsulated. At the same time serves as a proof of concept of the system: The physical object, bizarrely-shaped in nature, originated from the model itself and stored on the system, demonstrating the principle that these fictional drawings/patents have the potential of being (or not) realistically produced or replicated, in simple words the algorithm was in part responsible for the conception of its own interface.

This second element can also be considered as a navigation interface inside of this vast universe of never-before-created objects which is represented on a 2D display. When the controller generates the key via a QR-code we can access the third and final component which is the repository: a speculative archive —a patent office of sorts— that stores both the human and automatically generated drawings that are also described, named and taken; patented.

First Prototype was created using P5JS in combination with RunwayML, ESP32. Generator uses StyleGAN2 model running on Pytorch. Python. Controller uses a custom PCB, Raspberry Pi Zero, UDP protocol for live communication with the model. Python. Repository is HTML, Javascript, WebGL and Firebase.

The current state of the repository can be explored at: https://overloaded.supply

Project Page | Ignacio Pérez

Leave a Reply

Your email address will not be published. Required fields are marked *