Computational models of creativity have been inspired from psychological and cognitive studies of human creativity. Early efforts in artificial intelligence lead to computational creativity using models of analogical reasoning, emergence, reinforcement learning, and evolution. Recent developments in artificial intelligence using deep learning models from large datasets of words and images has enabled a new kind of computational creativity: one that is based on operations in a latent space. We can now measure the novelty and surprise of concepts and images in a latent space that can be the basis for encouraging curiosity and creativity. This trajectory in artificial intelligence as a way to model creativity, coupled with HCI approaches to supporting human creativity, has led to a third kind of creative system: co-creative systems. In a co-creative system, the computational system contributes to the creativity in the design or problem solving, encouraging curiosity and creativity in the human partner. In this talk, I will present the evolution of AI based computational creativity and the significance of new deep learning models for curiosity, novelty, and surprise. I will describe the Creative Sketching Partner that uses word embedding and image models from which a designer is presented with sketches that are a conceptual shift from the designer’s current sketch. The results of user studies show that such conceptual shifts inspire different kinds of creativity depending on how similar or different the system’s sketch is from the designer’s sketch.
Mary Lou Maher is Professor and Chair of Software and Information Systems at UNC Charlotte. She completed a Bachelor of Engineering at Columbia University and a MS and PhD at Carnegie Mellon University. She has been a Professor at Carnegie Mellon University, MIT, Columbia University, University of Maryland, and University of Sydney in Australia. She was a member of the Senior Executive Service at the National Science Foundation. Dr. Maher’s current research interests include: AI and cognitive models of creativity and curiosity, human centered design, and rethinking CS education.