AI Study
for performers, Live-AI, video, mobile app
LAYOUT:
ID: aistudy
GLOW: #EBFF57
≡
VIMEO BLOCK
≡
TEXT BLOCK
≡
TEXT BLOCK
Parallel Universe Access
(05:58 in the video)
≡
CODE BLOCK
≡
TEXT BLOCK
The audience engages with the performers through an app on their smartphones.
Using face detection, the app's camera overlays either AI-generated video personas onto the performers or robotic faces onto anyone visible within the camera frame.
Through this mediated view, the audience encounters parallel versions of the people present, creating a metaphorical experience that connects physical presence with artificial representation.≡
IMAGE BLOCK
≡
TEXT BLOCK
AI-Generated Videos
≡
VIMEO BLOCK
≡
VIMEO BLOCK
≡
TEXT BLOCK
I generated a series of videos based on the real performers’ profiles and that of Alan Turing. The videos function as metaphorical narratives around the concept of a parallel universe.
≡
TEXT BLOCK
Salon des Amateurs, Düsseldorf
Heterotopie
Year: 2025
Setting: any number of musicians
+ Live-AI
++ video and mobile application
Duration: ca. 30′ (various, open form)
Excerpt video: ca. 7'
AI Study explores the potential of contemporary AI models in the context of live interaction within a concert environment, as well as audience engagement. The piece consists of two sections: Live AI Session and Parallel Universe Access and also includes AI-generated videos.
The piece creates a shared environment in which performers and AI can engage in open interaction. To facilitate this environment, I developed an interface that enables voice-based communication between performers and a large language model (LLM), functioning as a kind of real-time composition workshop. In addition, a smartphone application allows audience members to participate directly and immediately during the performance.
≡
TEXT BLOCK
Live AI Session
(00:35 in the video)
≡
IMAGE BLOCK
≡
TEXT BLOCK
In the session, the performers on stage execute live performance instructions generated in real time by the AI acting as a composer.
At the beginning, one of the performers inputs a description of the current situation and a detailed set of criteria—established by the composer (myself)—that define what art and creativity mean. Notably, no description of the composer's own artistic practice is provided.
Based on these criteria, the AI first formulates a core artistic concept and subsequently generates a time-based composition. The session serves as a kind of composition workshop with real-time performance.
"Novelty is based on a unique structural or conceptual synthesis."
"We do not want a style of music."
The full criteria given to the AI can be found here: criteria
≡
IMAGE BLOCK
≡
TEXT BLOCK
The AI appears to take into account the immediate context of the culturally diverse group when formulating the core artistic concept. The resulting sequence of instructions is often highly unpredictable yet detailed, while at times remaining partially interpretable to the performers.
View the AI composition generated at the premiere: AI composition