Jul 25, 2017 - Jul 25, 2017
18:00 - 20:00
The Algebra of Artificial Intelligence
Newhaven Lecture Theatre, 15 South College Street
A public lecture by Daniel Murfet, University of Melbourne
It is often said that if we were to make contact with intelligence from another planet, we would communicate in the shared language of mathematics. What about the artificial intelligence that we are busy giving birth to right here on Earth? What kind of mathematics will we use to speak to our AIs?
The full answer to this question belongs to the future, but looking backwards we can trace many threads in the history of mathematics leading to the theory of artificial intelligence that we have today. This history offers us some insight into the kind of mathematics that we will use to communicate with AIs, both the "narrow" ones that we already have and the "general" ones when (and if) we figure out how to build them.
I will touch on some of these threads, including Gottfried Leibniz's unfulfilled dream of mechanical devices for performing calculations in a universal algebra of reason (in the 17th century!) and Norbert Wiener's ideas in the 1940's about control and communication between humans and machines. I will then follow these threads through to the present day, and explain some of the algebra involved in cutting edge artificial intelligence systems such as Neural Turing Machines, being developed in the UK by Google DeepMind, before speculating about other areas of algebra that might play a role in the future of AI.
The lecture will be followed by reception.