For this project I used data from the Metropolitan Museum of Art's dataset, GPT-2-simple, and the Twitter API to create a twitter bot that would post titles of Surrealist works of art that don't exist.

The process for this project is documented more thoroughly on my github page, but ultimately the workflow fell into three sections:

  1. Retrieving relevant titles from the dataset
  2. Training the GPT2 model on those titles
  3. Creating a twitter bot to post the generated titles

Ultimately I wanted to create a deliverable that would encourage people to engage playfully with art. Surrealism (and the movements it has inspired) encourages this kind play, reflection, and use of the reconstruction/détournement/transformation of material.