Distributed Creativity

November 2019


Language: Python

Overview

* Still an early exploration of ideas.

“Distributed Creativity“ is an experiment criticise the definition of creativity and how to approach it. The experiment runs as an artist workshop, in which the participants first learn the basics of Artificial Intelligence / machine learning and then driven by a set of guidance and rules to collaborate with other peer participants with the AI’s assistance. None individual participant nor the AI or the artist is driving the creative process solely but every of us is complementary. The contributions are distributed, so are the creation and the creative process.

Here the experiment with my students of Artificial Intelligence Arts class at IMA, NYU Shanghai:



How did I came with the Idea?

The idea of "Distributed Creativity" came from a fun exploration of BigGAN and some discussion on social networks. On a sunny autumn afternoon, I figured out how to interpolate bigGAN and obtain a smooth transformation from one generated image to another, all by a smooth walk in the latent space. I got some video like this:



Or from this vase to another!



I posted it on my Wechat, got the most questions about how did I generate the images in between and what kind of parameters I am using to make/guide the smooth transformation while I didn't control it at all. All my input to the system is the starting and ending point, all the rest are decided by the algorithm. That's how the started to think, the creative process is distributed, I, as a creator, didn't do more (important) work than the AI.

Inspired by collaborative art, I want to push the boundary further. What if I make the few-enough decisions more distributed and introduce in more "decision-nodes", here is a ruleset based collaborative art project I created in my AI Arts class at NYU Shanghai: Distributed Creativity.

And the rulesets we've been using:



More explorations on the way!

Source code for this website is available on github.