Can Computers Be Creative?

Ask Siri what time it is or ask Google for today's weather and you'll get what you need. Punch a mathematical expression into a calculator and you'll almost never be let down. Computers are incredibly precise and reliable when it comes to these types of tasks.

"Hey Siri, why doesn't my girlfriend like my paintings anymore?"

"Apple doesn't tell me everything you know."

On the other hand, there is a world of tasks computers are terrible at resolving. Why the stark difference? Will computers ever be able to understand art or be creative? Let's dig a little deeper.

Two Types of Problems

Let's divide our world of problems into two types. Firstly, there are problems that have a definitive answer, or multiple definitive answers. For example: Nine is divisible by which numbers? The answers are one, three and nine. The wonderful thing about these types of problems is that resolving them can be broken down into a simple sequence of steps, a procedure or a program. That's why computers are so great at solving them. You just run the program and it spits out the answers, often at a remarkably fast pace. It's incredible how sophisticated these algorithms, programs, have gotten and how powerful they can be. My favorite example of this type of ingenuity is IBM's Watson.

There is a second set of problems we deal with in our lives, and these problems have no definitive answer. They are entrenched in context, deal with changing variables or unknown factors. It is either more difficult or not useful to solve these problems by taking a prescribed set of steps to reach a conclusion. The solution is often contextual, transient, or mysterious unto itself. How to write an inspiring story or make a beautiful painting? How to choreograph a dance that will dazzle an audience? We come across these challenges in the arts but also in business. Which marketing campaign will succeed? Which logo best represents our brand? So far, computers have not been very successful in this realm. The realm of the creativity.

The Heartbeat of Creativity

When we look at this second set of challenges, and observe humans that are solving these types of problems (writers, musicians, marketeers), we find them leveraging their imagination. We find them using their creativity. So, what is Creativity? Creativity has a collection of definitions, many of which highlight two factors: Novelty and value. Something which is creative has an element of novelty, or newness, and an element of value, or utility.

When a musician writes a new song lyric or when an inventor sets out to design something she considers hundreds of possibilities before converging on an appropriate option. A marketing team generates a huge number of ideas before investing the time and resources to develop a campaign for its potential customers. The next time our musician writes or our inventor innovates, they will begin again by diverging on all possibilities before converging onto the most appropriate choice. This oscillation of divergence and convergence is what I like to call the "Heartbeat of Creativity." There is a huge volume of scholarly articles on this pattern of divergence and convergence with respect to the psychology of human creativity.

This pattern shows up in another place that is significant to our original question of whether computers can be creative: Evolution. In evolution divergence happens in the variance produced by sexual reproduction or mutation, then convergence happens when the environment selects the varieties most suitable for survival. Evolution, like the creative mind, is constantly diverging on possibilities and converging on the most appropriate options.

What About Computers?

Let's return to our original question: Can computers be creative?

I see hope. Pandora is a music website that curates music for you based on your and other users' preferences. Often times Pandora plays a song for you that you've never heard before (novelty) and that you're likely to enjoy (value). Choosing music that you will enjoy is certainly the type of challenge with an indefinite, transient and contextual solution. What about the creative process of divergence and convergence? That too, is beginning to make it into modern algorithms and programs. Take a look at the science behind Watson and see if you can spot the divergence and the convergence.

It seems to me that psychologists are beginning to better understand human creativity, and that engineers are beginning to learn to program it into computers. Perhaps the next great generation of artists will be made of silicon.