The Creativity Code: How AI is learning to write, paint and think

by Marcus du Sautoy

creativity is the drive to come up with something that is new and surprising and that has value.

To pass the Lovelace Test, an algorithm must originate a creative work of art such that the process is repeatable (i.e. it isn’t the result of a hardware error) and yet the programmer is unable to explain how the algorithm produced its output.

Exploratory creativity involves taking what is already there and exploring its outer edges, extending the limits of what is possible while remaining bound by the rules.

Boden believes that exploration accounts for 97 per cent of human creativity. This is the sort of creativity that computers excel at: pushing a pattern or set of rules to the extremes is perfect for a computational mechanism that can perform many more calculations than the human brain.

The second sort of creativity involves combination. Think of how an artist might take two completely different constructs and seek to combine them.

It is Boden’s third form of creativity that is the more mysterious and elusive, and that is transformational creativity.

Creativity is not an absolute but a relative activity. We are creative within our culture and frame of reference.

We have an awful habit of romanticising creative genius. The solitary artist working in isolation is frankly a myth. In most instances what looks like a step change is actually a continuous growth. Brian Eno talks about the idea of ‘scenius’, not genius, to acknowledge the community out of which creative intelligence often emerges. The American writer Joyce Carol Oates agrees: ‘Creative work, like scientific work, should be greeted as a communal effort – an attempt by an individual to give voice to many voices, an attempt to synthesize and explore and analyze.’

Unless you are prepared to fail, you will not take the risks that will allow you to break out and create something new.

But why would anyone want to use computers to create art? What is the motivation? Isn’t art meant to be an outpouring of the human code? Why get a computer to artificially generate art? Is it commercial? Are the creators trying to make money by pressing ‘print’ and running off endless new pieces of art? Or is this meant to be a new tool to extend our own creativity? Why do we as humans create art? Why is Richter’s work regarded as art while a book of Dulux colour samples is not? Do we even know what this thing we call art really is? Where did it all begin?

Art is ultimately an expression of human free will, and until computers have their own version of this, art created by a computer will always be traceable back to a human desire to create.

for many using computers in their art this is simply about a new tool. We never regarded a camera as being creative but rather as allowing a new creativity in humans.

Eran Kahana, an intellectual-property lawyer at Maslon LLP and a fellow at Stanford Law School, explains that the reason why IP laws exist is to ‘prevent others from using it and enabling the owner to generate a benefit. An AI doesn’t have any of those needs. AI is a tool to generate those kinds of content.’ What if AI creates a piece of art in the style of a living artist? It’s likely that the programmer might be sued for copyright infringement but it’s a very grey area. Inspiration and imitation are central to the artistic process. Where is the line between your own creation and copying someone else’s?

Research by neuroscientists has discovered that, like the algorithms driving the Generative Adversarial Networks at Google Brain, our own brains have two competing systems at play. One is an exhibitionist urge to make things. To create. To express. The other system is an inhibitor, the critical alter ego that casts doubt on our ideas, that questions and criticises them. We need a very careful balance of both in order to venture into the new. A creative thought needs to be balanced with a feedback loop which evaluates the thought so that it can be refined and generated again.

David Cope first turned to algorithms out of desperation. He’d been commissioned to write a new opera and was procrastinating, unable to commit notes to stave. But then he had an idea. He remembered how Ada Lovelace had speculated that ‘the engine might compose elaborate and scientific pieces of music of any degree of complexity or extent’ and decided to explore her idea. He started experimenting by feeding punch cards into an IBM computer (this was the early 1980s). Notes appeared as output. His early experiments, he later admitted, were truly dreadful. But he persevered, heading off to Stanford to do a course in computer music. With the deadline for his opera commission fast approaching, he decided to put his computing skills to the test. If he could get an algorithm to understand his compositional style, then whenever he got stuck and didn’t know how to proceed, the algorithm could make a suggestion that would be compatible with his particular way of composing. Even if the algorithm suggested something Cope considered absurd, it would at least help him understand what might have been a better choice. The algorithm would act as catalyst to spur his creativity. Cope called his new concept ‘Experiments in Musical Intelligence’, or EMI for short. The alter ego composer that began to emerge from these algorithmic experiments would later be called Emmy, partly to avoid confusing the project with the British recording label EMI, partly to give the algorithm a more human name.

On another occasion, when Hofstadter played two pieces, one by Chopin and the other a Chopin-like piece composed by Emmy, an audience made up of many composers and musical theorists were duped into believing that the computer-generated piece was the real one.

The new ideas of machine learning I think challenge many of the traditional arguments about why machines can never be creative.