Creative Cookies through Evolution
By Christina Irakleous, Nikki Rademaker, and Yanna Smid
This project was part of the MSc Media Technology course Computational Creativity. The assignment was to create a system that could generate new cookie or cake recipes based on an inspiring set, using a genetic algorithm. In the end, we had to create a small cookbook presenting several of the recipes made by our system. Generative AI could be used to illustrate and expand the recipes, but the generator itself had to be our own.
The project taught us how to use a genetic algorithm in a creative context. We also learned how AI can support but not replace human creativity when it comes to presentation and communication.
We created a Python-based generator called the Magical Cookie Creator. It uses a genetic algorithm to evolve new cookie recipes by combining ingredients from 43 real-world cookie and cupcake recipes. The system rewards variety and encourages the use of flour. It selects and mutates recipes over generations to create more original combinations.
To present the output, we selected 7 recipes that stood out and used ChatGPT to generate clear baking instructions. DALL·E was used to visualize what each cookie might look like. We collected everything into a mini digital cookbook, which we designed ourselves using Canva. Each recipe comes with a title, ingredients, baking steps, and a generated image.
You can flip through our cookie recipe cookbook here. It includes cookies like "Rainbow Muffin Unicorn Dip" and "Velvet Apple Pretzel Mints". These are the kind of strange and surprising recipes our algorithm would dream up.