This page will allow you to generate your own instructions for performance art pieces in the style of our current exhibit Embodying the Algorithm using GPT-3’s DaVinci model. Each time you refresh this page, a new performance that has been previously generated using GPT-3 will be displayed. If you would like to record and submit your own performance based on one of the generated texts, please see the instructions on our Submissions page.

Refresh Page to Generate a New Performance

Instructions for a performance by GPT-3:

An idea for a performance art piece: a single performer, a man, sits in the middle of a circle of chairs, which are occupied by people. The man sits in one of the chairs, and the others sit in the other chairs. The whole scene is lit by a single lightbulb, which hangs from the ceiling just behind the seated man.

The man is instructed to say whatever comes to his mind. The others are also instructed to say whatever comes to their minds. Their instructions are to say the first thing that comes to their minds, and they are not allowed to say anything that has already been said. They must also not repeat the first thing that comes to their minds unless they are sure that the first thing has already been said. If a person repeats the first thing that comes to their mind, they are out of the performance piece.