Advanced Generative Art and Computational Creativity

This course proposes a deepened survey of current practices in generative arts and computational creativity with an emphasis on the formal paradigms and algorithms used for generation. In this advanced class, we study how evolutionary computing, neural networks, and procedural generation can produce novel and valuable artifacts. We survey advances in search-based methods and procedural generation. We look at how to formalize aesthetic measures and learn how creative systems can be evaluated.

We illustrate how these algorithms have been used in numerous examples of past and current productions in visual art, new media, music, poetry, literature, design, architecture, games, moving images, and robot-art. Students get to practice these algorithms first hand and develop new generative pieces through assignments and projects in MAX.

Finally, we discuss the societal and ethical implications of the automation of creative tasks, from the fear of artificial intelligence to the algorithmic bias, and from the most technophobic visions to the most technophilic ideals.

• Societal and Philosophical Perspectives
• To conclude this class, we put generative practices in the more general context of media art. We discuss the fear of automation, and the algorithmic bias. We then present the underlying philosophical debate between technophobia and technophilia and discuss its implications on the relationship between art and science.
• Evolutionary Computing and Genetic Algorithms
• After a brief introduction to this second part on the topic of generative art and computational creativity, we introduce evolutionary computing and learn how genetic algorithms can be used to evolve new artifacts in visual art and music.
• Evaluation Methods for Computational Creativity
• We learned how to develop generative systems for a wide variety of creative tasks, but how good are they? In this session, we cover both informal and formal evaluation methods. We also introduce live coding and discuss the possible bias against computational creativity.
• Search-based Approaches to Creativity
• Most creative tasks can be framed as a search problem. This session details advances in procedural content generation for games, and story generation. We also review generative methods used for moving images, dance, choreography, and survey progresses in art making robots.
• Genetic Programming and Evolutionary Ecosystems
• In this session, we study how genetic programming is used to breed programs to that generate new artifacts in design and architecture and how evolutionary forces can be used to breed behavior and populations of agents in ecosystemic artworks.
• Artificial Neural Networks and Deep Learning
• This session introduces artificial neural network, and present the perceptron, and multi-layer feed-forward network. Artistic applications of self-organizing maps, neuro-evolution and deep learning are reviewed and discussed.