Thank you for your participation in the algorithmic ceremony Artificial Awakening.

The following content of this performance were made by an Artificial Intelligence (A.I.) in collaboration with humans. These are the various A.I.’s that you have been in contact with during the ceremony:

• The script and performance concept was co-written with Open-AI’s GPT-3 text generation model with 175 billion parameters and trained on text from the internet.
• The A.I.’s voice was generated using a state-of-the-art neural text-to-speech machine learning model.
• The pulsating AI Hallucination room in virtual reality was generated from a model trained on 40,000 Google Street View 360 degree images using Nvidia’s StyleGAN2 architecture.
• The full body avatar of the virtual performer was generated via a styleGAN Machine Learning Model that was fed with 12 key image frames from Artbreeder Library, that was then interpolated to 1500 images creating a 1 min. animation wrapped on a 3D body.
• The Face Masks in virtual reality was made from faces of people that does not exist using StyleGAN2 model trained on 70,000 human faces and the 3D Dense Facial Alignment model from Jianzhu Guo et al. that converts 2D images of human faces into 3D.
• Videos of participants on the wall of faces at the end of the performance was made by projecting them into a “latent space” vector using a StyleGAN2 model trained on 70,000 human faces and applying latent vector directions to make them age.
• The choir singing at the last stage was written and sung by OpenAI’s Jukebox AI model trained on 1.2 million songs.

Your contribution will now be part of the future experience and refinement of the ceremony.

Credits
Director & Writer: Jakob la Cour
Producer: Mads Damsbo
VR Artist: Cenk Güzeliş
Performer: Eja Due
Guide: Nina Runa Essendrop & Anton Sager
Scenographer: Cæcilie Tørnsø
AI Artist: Valdemar Danry
Composer: Fabian Lanzmaier
Technical Artist: Carl Emil Carlsen
Developer: Burkart Schwaighofer & Balder Brüsch
Creative Technologist: Jeppe Etzerodt
Machine Learning Engineer: Esbern Torgard Kaspersen
Researcher: Linnea Bjerregaard
Head of Production: Jacob Weitze Mylund

The project is generously supported by
The Danish Arts Foundation