Intercative light-sound art installation with AI



The installation affects a person at his instinctive level, just as a person affects the installation. The viewer and the work itself, which uses the capabilities of artificial intelligence, are in a state of contradiction with each other, since one poses a potential danger to the other.

The installation learns to defend itself keeping people away from it, but attracting them if they are at a safe distance. The neural network determines the necessary dynamics and intensity of the lamp glow, as well as the volume and nature of the generated sound. The obtained data on the influence of light and sound on the distance to the participants will again be used as material for training the model, so the installation develops and reveals more and more effective and perfect ways of influencing people for self-defense.

To understand complex things, we need to begin with a simple one. The instinct of self-preservation is what unites all living beings. We want to consider this ancient mechanism of our life in a new context - the context of artificial life and artificial intelligent. It's known, machine learning models are somewhat similar to the structure of the nervous system of complex organisms. From the very beginning, humans created AI for interaction and use it as a tool. Therefore, we cannot call a separate learning system "a subject", since this model is not isolated and completely depends on its creator or user. By endowing an artificial system with the instinct of self-preservation, so that it learns to survive and avoid danger, we implement this system in a separate subject. Leaving out the question of awareness of its existence, at least this system is trying to protect its existence.

Perhaps, exploring and observing a system that has one of the basic instincts we will learn something new not only about the features of AI, but also about ourselves. If you look at familiar things in a different light, you always start to notice new details.

Equipment: 4kW lamps, IR Lidar, Raspberry PI, Arduino, 4 Speakers 

Software: TensorFlow, Python, OpenAI Gym, Puredata


·        Chronus Art Center, Shanghai, China

·        “Signal” festival, Nikola-Lenivec