Many believe that empathy is the realm of animals. So only people can create, right? After all, creativity is one area that relies on empathy to work. Think of your favourite song, book, or any memorable part of a film or TV series that you remember. Those resonated with you because they made you feel something. That’s what sets us apart. Or at least, it used to.
What if I told you that machines are capable of creativity now? With Google’s DeepMind able to dream – and use those dreams to accelerate its rate of learning – we’re on the precipice of a new era in creativity.
We have employed the help of machines to make our creative work easier and faster to produce for many years. More recently, we began to use real data and algorithmic thinking to produce work which is vastly more scalable, customisable and personal than ever before. A great example of this is the last two rebrands for MIT Media Labs undertaken by design firm Pentagram. They created several thousand permutations of their logo to give all faculties their own icon. The idea is that as no two people are alike, the brand should be equally as unique. The result is striking and highly individual, and something that until fairly recently would not have been feasible.
Mainstream ad shops are already embracing AI and machine learning with campaigns from Toyota, Coca-Cola and Burger King. Similarly to MIT’s work, Toyota used AI to create thousands of variant headlines that tapped into insights hidden deep within research. Meanwhile Coca-Cola is using AI-powered chatbots built in to vending machines to deliver unique customer experiences. More recently, Burger King had an AI write entire scripts for TV spots. The result is bizarre but it’s memorable and kind of funny. Maybe the machine has learned that’s what we really want…
But forget branding and marketing. The real magic is happening in gaming, music and film. Procedural generation in gaming (Diablo’s random corridors, the virtually limitless worlds of No Man’s Sky) has been utilised for a long time, but in 2016, Sunspring was released. This short film written entirely by Benjamin…an AI. Sure, it was then filmed by humans, but this was an invented story about three people connected through a love triangle – written by a machine.
On the music front, Sony CSL Research Laboratory gave an AI thousands of pages of sheet music and instructed it to write a song. The result was Daddy’s Car. Very much in the vein of The Beatles and The Beach Boys it’s… NOT Pet Sounds. However, it’s also far better than you might expect. An Australian company, Popgun, is even using AI to develop software that makes writing music easier, while Google is using an AI called Duet to jam with people. Eventually, I imagine both of these will use the knowledge they gain to create their own music.
HAL 9000, 2010 Space Odyssey
For AI to truly learn to be creative it will need to learn the rules then break them, the same way we do. Currently, it is still reliant on humans giving stimulus or reference in order to create, ensuring that our jobs are safe for the time being. However, as they learn from this information and feedback, they will quickly be able to create outside the parameters we set. That will be when AI gets taken seriously as a creator.
For now, it’s up to us to guide them in the right direction – as Microsoft found out the hard way with its racist chatbot – so that we can work with emerging AI. Ultimately, all of this technology is part of the democratisation of creativity, allowing nascent creatives to realise their dreams. Like much of human history, we tend to be afraid of new things only to discover that (in most cases) they actually improve things.
We don’t need to be afraid of the machines just yet. But we do need to start thinking in ways that will allow us to harness their power. Until AI develops true empathy, we will be stuck aiming to work together but not quite understanding each other. Once computers learn to feel what we feel, and we learn to embrace it, we will open a doorway to a new creative revolution.