Befriending the Algorithm
How to turn your computer into the next major featuring artist
“The current state of creative AI is very very imperfect, it doesn’t quite capture the thing you’re looking for, but in those failures there’s a lot of interesting stuff.”
Exploring how collaboration between human and machine-like entities (in particular neural networks) for creative purposes can produce new and unexpected outcomes and what that can mean for design practice.
To demonstrate what kind of design artifacts this collaboration with AI can produce.
To try to demystify the stereotypes about creativity related AI usage .
- Choosing a machine learning model to collaborate with. The chosen model was a Generative Adversarial Network (GAN).
- Choosing what kind of GAN to collaborate with. Styletransfer is the first option, it can produce an output based on a certain aesthetic given by an input and apply it to anything.
- Establishing a purpose for training the neural network. What kind of dataset will be compiled?
- Producing a catalog based on the results with the purpose of documenting the various outcomes and how they were produced and also to reflect about the experience.
The aim is to induce reflection about what roles can AI take when associated with creative practices.
Collaboration can take many forms and not just human-like ones. Artificial Intelligence, especially neural networks, can do several tasks that humans just can’t. Collaborating creatively with these entities means that the two parties bring different assets to the table and what is interesting is seeing what can come out of it.
The purpose is not to claim that creative collaboration with AI is the future or the way to go, as neural networks are still really early in development, nor to claim AI as the destruction of creativity. But simply to say that the option is still there and under-explored, and that it can open up a lot of paths.
This project is an exploration of a few of those paths.