Art and AI Family Tree

By Atreya Ray

The intersection of Art and AI currently has tools which rely heavily on data produced by humans. This is because results produced by ML tools usually yield results based on the training data, which in this case is artwork by humans. The idea of this artwork would be to compose several ML models on top of each other to create layers of abstraction, distancing the final product further and further away from human made art and making it more of a true unique piece of art made by the AI itself.

This “family tree” shows the outputs produced when several AI tools are stacked on top of each other in different permutations. Only the initial prompt “Art and AI” has been used to generate different sorts of art pieces. Text , text to image, image to text generators and style transfer tools have been used to change the medium of the art and to show some of the art pieces in more classical styles.

Tools Used: Runway ML 

  • AttnGAN: Fine-Grained Text to Image Generation with Attentional Generative Adversarial Networks
  • XLNet Text Generator
  • Neural Style Model
  • Adaptive Style Transfer Tool
  • im2text Generator
  • Film-Script