top of page


A modern dance piece utilizing machine learning to investigate how modern dance can be challenged and enriched through the integration of machine learning

Articulating contemporary tensions between man and machine, the performance Centaur, is a modern dance piece that utilizes machine learning technology in an exploration into the consequences of modern day artificial intelligence and the power of technology in a clash between creators versus their creation.

Centaur is developed by critically acclaimed artistic director and choreographer Pontus Lidberg in collaboration with AI-artist Cecilie Waagner-Falkenstrøm and Ryoji Ikeda, Japan’s leading electronic composer and audio-visual artist.

Inspired by the half-horse half-human creature from Greek mythology, the concept of the modern Centaur is also the name of the computer-science concept relating to the elevation of natural human qualities through the combination of both artificial and human intelligence.

To this effect the artwork scrutinizes what happens when the creative power of artificial intelligence entangles with a set of 9 dancers each night through a machine learning based installation live on stage. Throughout the performance the dancers encounter a unique ARTificial intelligence, programmed exclusively for the performance by artist Cecilie Waagner Falkenstrøm and her team of software developers and data scientists. The ARTificial intelligence is both co-creator and participant in the performance, as it influences choreographic composition based on various data sets such as planetary movements, swarm technology and the dancers’ previous movements as they have been tracked and collected throughout the creative process. As a performer, the artificial intelligence can simulate consciousness, emotion and intention as it interacts with the dancers on stage. This means that each performance is a unique and unpredictable event — a neat allegory to our human relationship with technology.

Throughout the process and rehearsals, the technology and humans have been working closely together. Human created compositions have been deconstructed by the machine and vice versa in an ongoing process of creation with the final result being something entirely new that neither the choreographer nor the artificial intelligence algorithm could have anticipated. There is no divisible point where the human creation ends and the machine takes over – both are entangled.

AI-based Modern Dance Performance:

Artistic team:
Choreography: Pontus Lidberg
Artificial Intelligence installation: Cecilie Waagner Falkenstrøm
Audio and Visual Design: Ryoji Ikeda
Computer Programming for Visuals: Tomonaga Tokuyama
Light Design: Raphael Frisenvænge Solholm
Costume Design: Rachel Quarmby-Spadaccini
Dramaturge: Adrian Guo Silver
Music: Franz Schubert, Ryoji Ikeda, Giacomo Puccini
Computer Programming for AI: ARTificial Mind & Technical University of Denmark – DTU Compute Department
Development: The Center for Ballet and the Arts at New York University
Co-production: Danish Dance Theatre co-produced with Oriente Occidente, Théâtre National de Chaillot and The Royal Danish Theatre

Tech development for Artificial Intelligence:
ARTificial Mind and Technical University of Denmark – DTU Compute Department


Awards & Grants:

Cecilie Waagner Falkenstrøm/ARTificial Minds contribution to the performance is the ARTificial intelligence.

Artificial intelligence technology:
In the performance we utilize a range of machine learning technologies to give life on stage to the artificial intelligence-based centaur. Recurrent Neural Networks (LSTM) have been used to generate an infinite number of new compositions and choreographically elements based upon a dataset of sensor-tracking of dancers’ performing choreographic compositions created by Pontus Lidberg. Swarm intelligence has enabled a set of rule-based interactions between the individual dancers of the company leading to the emergence of an intelligent overall group behavior, unknown to the individual agents. Reinforcement Learning (Genetic Algorithm) is used to enable the algorithm to strive toward creating a composition based on certain goals, such as like symmetry/asymmetry, quantity of dancers on stage and collisions avoidance. Natural Language Processing (Transformer ML-model) enables the artificial intelligence to recompose dramaturgical literature into new poetic narratives. The AI’s machine learning-based humanized voice (Speech Synthesis) allows it to give live instructions to the dancers each night.

The algorithms have been fed information from a myriad datasets ranging from planetary movements and Greek tragedy to sensor tracking of dancers. Filled with knowledge, the different algorithms utilized in the performance enables the AI protagonist to compose a dance in real time by instructing the dancers as to which poetic movement phrases to work with, responding to what dancers chose to accept and creating various layers of movement in response.

Centaur, Musikhuset Aarhus, 21. sep 2022
Centaur, Teatros del Canal, Madrid, 1. april - 3. april 2022
Centaur, Royal Danish Theatre, Copenhagen, Oct 21st – Oct 25th 2021
Centaur, Théâtre National de Chaillot, Paris, Jan 28th – Jan 30th 2021 (cancelled due to the Covid-19 pandemic)
Centaur, Wolfsburg, Germany, Sep 29th 2020
Centaur, Festival Oriente-Occidente, Italy, Sep 3rd – Sep 4th 2020

Thanks to:
Augustinusfonden, Overretssagfører L. Zeuthens Mindelegat, William Demant Fonden, Knud Højgaards Fond, Beckett-Fonden, Jyllands-Postens Fond


bottom of page