AI and Creativity
Machine learning (ML) is transforming our reality. Today’s models deeply learn about their input domains and reach beyond-human capabilities on a growing list of tasks. Beyond classification and decision making, can AI be creative?
While learning to predict, many ML models construct a view of their input worlds that allow them to also generate new input data – they are generative models. Generative adversarial networks (GANs) is one class of machine learning algorithms that are being increasingly used in creative domains to generate novel instances of music, visual art, prose, and more. In this hands-on class, students will use generative ML models to create digital art. Examples include “painting” and “sketching”, “writing” a poem or a whole (fake-)news article, transfer artistic styles to create new visual art, generate faces that don’t exist, hallucinate structure out of noise, generate 3D models, generate videos, and “compose” music. The class will focus on hands-on projects, some do not require any coding experience, and others involve simple coding in Python.
The class will discuss how autonomous creative AI systems can collaborate with humans to foster creativity in humans. Lastly, the class will cover critical discussions around the ethical implications of generative modeling such as misinformation and fake media through unplugged activities and debates.
Students will learn how machine learning models are used to generate information in multiple media (text, sound, picture, 3D geometry). By the end of the course, participants will be able to apply these tools to create their own project. The course is designed for persons without prior experience in machine learning or coding.
This course is appropriate for everyone–from computer scientist to journalist to concerned citizen. It is appropriate for individuals at any company—from global corporations to small start-ups—that develops AI system capabilities or is a user/buyer of AI systems. Participants are welcome from a wide range of industries including energy, consumer AI products and services, aerospace, transportation, robotics systems, finance, national security, and health. Participants interested in the implications of creative AI systems, and how AI can help people be creative, are especially encouraged to participate.
Cynthia Breazeal, Pattie Maes, Zach Lieberman, Ishwarya Ananthabhotla, Safinah Ali, Daniella DiPaola
Introduction to Creative Machine Learning. How can machines help us create art?
Generative Modeling Techniques: Generative Adversarial Networks, Variational Auto-Encoders, Recurrent Neural Networks
Working lunch: discussion
How do GANs work? Interactive web tools to explore examples of GANs
Deeper into Generative Adversarial Networks for generating text, sketches, images, videos, music
Working lunch: discussion and brainstorm: how can we apply these ideas to students’ areas?
Human-AI co-creativity. Can AI inspire creativity in humans? Interactive tools that foster creativity
Ethics of GANs
Working lunch: discussion and brainstorm: ethics of generated media
Generation and spread of misinformation
There are no pre-requisites to this course. This is an introductory course. Prior programming experience is helpful but not necessary.