by Jake, October 7th, 2019
Can a machine learn to improvise jazz? Is it possible for a piece of code to write an inspiring poem? Will a computer ever paint a masterpiece? Marcus du Sautoy, professor of mathematics at the University of Oxford, came to ISP this week to discuss some of these questions with the grade 12 students in relation to their Theory of Knowledge (TOK) course.
Living in an age when machine learning and artificial intelligence are becoming ubiquitous in the technologies used in daily life, we can’t help but ask whether computers could extend their abilities from just being able to perform calculations to synthesizing creative work. It has been known for a while that performing calculations is something computers excel at; they can multiply long strings of numbers faster than any human can and find prime numbers with millions of digits. The ability to synthesize creative work, on the other hand, is something most people would consider a human ability. Surely only conscious humans could possess the extraordinary ability to compose symphonies, write compelling novels and paint impressionist artwork?
In his talk, Professor du Sautoy gives us a glimpse into the world of artificial intelligence and the progress it has made in discovering the so-called “creativity code”. One of the most interesting examples of computer creativity he provides is that of AlphaGo, a computer program developed to play the complex board game Go. In the game of Go, players take turns placing black and white stones on a 19-by-19 board; the goal is to surround the opponent’s stones. Not only does this game require a lot of strategy and creativity, but it also relies heavily on intuition from experience. It was initially believed that this game was too complex for a computer program to master at a high level. However, AlphaGo was a special machine learning program that played various versions of itself
thousands of times, each time learning from its mistakes and thus morphing its code.
Eventually, this self-learning program became skilled enough to challenge the Go world champion and win with a score of 4 to 1. Interestingly, early on in one of these games, AlphaGo made what at the time was thought to be a critical mistake. It placed a stone 5 ranks from the edge of the board, a move that defied one of the very basic principles of Go playing. To everyone’s surprise though, this move happened to be crucial to AlphaGo’s win late in the game. This begs the question, “Can computer creativity surpass human creativity? Is the human way to do things the optimal way to do things?” Perhaps some of the things humans consider to be at an optimal peak are, in reality, just small hills among a sea of mountains.
In an interactive segment called “Bot or Not”, Professor du Sautoy showed the audience two works of art, one made by artificial intelligence and one made by humans, and conducted a live survey to see whether the grade 12 students could distinguish between the two. Whether it was Rembrandt paintings, abstract art or poems, it was quite clear that distinguishing between works of
humans and works of computers was very difficult. In fact, as professor du Sautoy puts it, the half-divided pie charts were reminiscent of Brexit votes!
In his engaging talk, professor du Sautoy explored the creative abilities of artificial intelligence. Machines and codes that can emulate human art and expression at an indistinguishable level raise a question that lies at the heart of TOK. Is creativity limited to conscious beings? As exciting as this topic is, it is impossible to pack everything into a single talk and we were extremely lucky to have an esteemed mathematician speak at ISP. Not only is professor du Sautoy a frontier researcher within fields of group theory and number theory, but, as a Simonyi Professor for the Public Understanding of Science, he is also central to the popularization of Math and Science. As an enthusiast in these subjects, I derive a significant amount of my sleep deprivation from watching lectures of thinkers such as Marcus du Sautoy. To be able to personally meet an idol that I have only been able to see through a screen, to discuss topics of mutual interest and to have him sign my calculator was an unforgettable experience!
#ISPresearch #ISPcreativity #ISParisIB
- artificial intelligence
- machine learning
- theory of knowledge