The Bad Idea Machine

Machine Learning

For my final project at Flatiron School, I decided to explore machine learning. I think a really exciting area of machine learning is to augment the human creative process – essentially giving us more ideas from which to filter and then develop "good" ideas. This site uses a GPT-2 text-synthesizing neural network to generate "pitches" for ideas, and randomly pairs them with images and brands.

Check it out:

I envision a use for this in the advertising industry, allowing teams to come up with unexpected ideas that they can refine and propose to their clients. These ideas are all "bad" initially, as they are raw and unrefined, but a creative team can use them as a jumping-off point.

The backend is a Ruby on Rails API, running on Heroku with PostgreSQL, and the frontend is React, deployed to Netlify.

Login Screen
Campaign Idea for Dr. Pepper
Campaign Idea for Ford
Campaign Idea for Tide

Web Portfolio

Film Portfolio