Cycle to Learn | Breath-Driven Interactive AI Art Installation

The knowledge in artificial intelligence emerges from the prior training process, where the parameters in the models repeatedly optimize themselves through iterations of forward and backward propagation. The unseen endeavors of these cycles fuel the accomplishment of the tools that are ultimately presented to the users. By penetrating the phenomenological results that we are accustomed to, we critically reflect on the causality and materiality entailed in this cyclical process. Just as AI models refine their understanding through countless cycles of data processing, human life relies on the rhythmic cycle of breath. This installation draws parallels between these two forms of intelligence, inviting participants to reflect on the shared cyclical nature of both. In the exploratory work, Cycle to Learn (2024), we adopted real-time AI-powered generative tools to examine the rarely mentioned prior process of emergence, together with the participants in an interactive audio-visual installation. The interaction design, which detects the unconscious cyclical breathing motion of the viewers, is also an experimental attempt in its instructional utterance and aims for deep embodiment and interrogative thoughts on the emergence of human intelligence and the relationship of our understanding to the knowledge and laws of the world through observations of periodicity.

Credit

Co-directed by Han Zhang & Mingyong Cheng

Sound Artist: Han Zhang

Visual Artist: Mingyong Cheng

‍