No AI was used to create content. Instead audience members, on-site or remote, whispered into their device creating a virtual “wind” that shaped how the environment looked and reacted to the performance. ChatGPT analyzed speech patterns focusing on context, tone and complexities. This information was mapped to generative visuals in Touchdesigner, adjusting the forest rendering in real-time.
Combined, these techniques resulted in an ambient proof-of-concept performance that occurred over several hours, allowing audiences to come and go for durations of their choice as the piece evolved over time.