Advanced Block III

An Introduction on Creative AI

Creativity is often considered the hallmark of human intelligence. The ability to use knowledge, imagination and to produce something that did not exist before has been seen as an incredible quality throughout history, so that the major innovators in any fields, both artistic and scientific, are commonly described by the same word: genius.

In this talk we try to define what is creativity and the key ingredients of a creative solution. We analyze why the majority of artificial intelligence models are not creative and the difficulties to instill creativity into them. After that generative models are discussed, with a particular emphasis on generative adversarial networks (GANs) and how they solve the problem of generating data in a creative manner relying on social interactions. After that some typical creative tasks resolved using GANs (generation of paintings, music, etc.) are discussed, with a focus on the so-called creative adversarial networks (CANs).

In the final part of the talk we'll explore creative models not based on machine learning, and their usage in science, engineering and design. For more questions, feel free to contact research@aiforpeople.org.

About the lecturer:

  • Gabriele Graffieti

  • He is the Head of AI Research at AI for People

  • Interested in: Generative models, continual learning and bio-inspired learning models

  • Currently, he is a PhD student in Data Science and Computation at the University of Bologna

  • For more: find Gabriele on LinkedIn