Mimicry

Ziwei Wu

Mimicry is a multi-screen video installation that explores evolutionary adaptation through computer algorithms, drawing parallels between biological mimicry and human society. In nature, mimicry is an adaptive strategy where species evolve to resemble their surroundings or other organisms for survival. Similarly, as Walter Lippmann described in Public Opinion, people construct subjective, biased mental images of the world, forming a pseudo-environment that shapes their perceptions and behaviors.

The installation captures real-time footage of plants, while algorithms generate and evolve virtual insects that attempt to blend seamlessly into the recorded background. Through a process of simulated breeding, selection, and mutation, these digital organisms adapt over time, visualizing an ongoing evolutionary cycle. The installation’s structure pays homage to Nam June Paik’s TV Garden, which envisioned a future where technology and nature coexist.

In the latest version, audiences actively participate by altering news text, which is then processed by an AI diffusion model to synthesize an artificial environment. This new environment influences the virtual insects’ mimicry process, guided by a Generative Adversarial Network (GAN). The GAN’s generator creates insect patterns that resemble the environment, while the discriminator, acting as a predator, attempts to differentiate them, driving an adversarial evolution.

Project Collaborators:

General concept, original idea, and research: Ziwei Wu Installation, sound, and visual design: Ziwei Wu Genetic algorithm implementation: Lingdong Huang AI and ALife system design: Ziwei Wu, Xiyu Tian and Wanchao Su AI and ALife system implementation: Xiyu Tian and Wanchaosu This project was researched, developed, and produced at the Hong Kong University of Science and Technology, Hong Kong University of Science and Technology (Guangzhou), and Monash University. With support from: Exhibition support by Lumen Art Project; exhibition space support by T33; academic support by Prof. Kang Zhang and Prof. Huamin Qu. Special thanks to Xinyu Ma, Kenny Ma and Prof. Jon McCormack.