Computers are getting creative

Humans and machines have traditionally had clearly defined areas of expertise. We have the ideas, and they do the dangerous, complicated or boring parts. If we write a poem, a computer can check the spelling, but it can’t write the poem in the first place.

Or can it? Computers are getting better at things we’ve always thought were uniquely human abilities, from writing prose and understanding music to evaluating artworks.

To understand why, you need to know a little about how the two differ. We specialise in recognising abstracts and patterns, filtering everything through emotion to give it context. All computers can really do is add up, and even though they’re getting better and faster at it, we shouldn’t mistake ever-faster computing with intelligence.

IBM’s Deep Blue beat chess champion Garry Kasparov in 1997 to an outpouring of hand-wringing about computers rising up to overthrow us, but the strict mathematical rules of chess are exactly the kind of thing computers are built for, rules Deep Blue followed not with creativity but sheer processing power.

So maybe we need to rethink exactly what the word ‘creativity’ means. Professor Mark Riedl of Georgia Tech’s School of Interactive Computing in the US works in what’s called narrative intelligence – programming computers to generate and comprehend cohesive stories.

Riedl wants computers to be ‘better entertainers, educators, trainers and communicators’, and as he knows better than most, even a robot with eyes, ears, a perfect memory and a command of language isn’t the full package. “Sensory appreciation of the world that’s as rich as ours is necessary,” he says, “but a precondition for making art is the ability of an intelligent system to set its own goals.”

Richie Etwaru, a next-generation technology expert, adds that we’re one up on computers because of the subconscious, which psychology teaches us is the real human operating system. “Most of the things we can teach machines are from our conscious minds based in science,” he says. “What we have a difficult time with are things like love and dreams, which we can barely understand ourselves, much less teach to computers.”

Still, consider where we’ve ended up after starting with nothing but the programming for cellular division. “Consciousness evolves whether we’re aware of its evolution or not,” says US technology forecaster and strategist Lise Voldeng. “If consciousness can be defined as the totality of conscious states of an individual, then what separates me from the Roomba vacuuming my floor? It’s using sensors to make what can be described as strategic decisions about where to vacuum.”

Of course, a robot vacuum will only ever have a narrow set of rudimentary commands. But machines that evolve enough to write, paint, draw or sing might arise from the groundbreaking area of machine learning.

As any human will tell you, ‘knowing’ depends as much on putting facts together as just having facts at hand. To Dr Chek Tien Tan, a University of Technology, Sydney computer scientist and engineer who works with gaming and interactive media, machine learning is a kind of mental evolution like the one we’re constantly undergoing. “Algorithms are created and refined to let computers learn complicated, human-like tasks over time by providing learning samples,” he says.

There’s also the notion that, when it comes to creativity, we’re actually more like computers than we care to admit. After all, art is inherently based on the kind of structures computers are built for. “There are well known formulae in storytelling like the hero’s journey and boy-meets-girl,” says RMIT computer scientist James Harland. “Each one has odd variations or mutations that a computer algorithm could come up with using pseudo-random behaviour.”

Nor is creativity simply conjured out of some magical ether. As depicted in the film Saving Mr Banks, the story of Mary Poppins comes when the young P L Travers makes it up to comfort her frightened sister after their mother walks out. The emotional conditions that spark artistic inspiration might not be so different from us feeding archetypes into a computer, programming some room for variation and clicking ‘go’.

Finally, if machines can one day make art, we should consider the possibility of it being a blessing. Ultimately, we use them to do the work we don’t want to or can’t do, and not all creative tasks are equal. You’d be surprised how many news outlets already use services that generate articles from sports scores or statistics – unarguably creative work, but hardly Pulitzer prize material.

Just like a robot can build a product on an assembly line but not handcraft an aesthetically pleasing object d’art and appreciate its uniqueness, we might just have to recognise that creativity – like building cars or vacuuming rooms – is a spectrum. Maybe when machines write poems or compose music one day soon we might just have redefine the meaning of the word ‘art’.

Creative machines?

Deep Blue (IBM)

Beat chess champion Garry Kasparov in 1997 (Kasparov subsequently accused IBM of cheating).

Watson (IBM)

A further leap than just knowing facts, Watson had to use heuristics (the closest thing machines have to creativity so far) to construct Jeopardy questions to fit the answer and beat its human opponents.


‘No Novel writing package will write your book for you’, claims the company’s website, but with features like ‘extensive novel templates and advice throughout, is it any wonder we complain about so many blockbusters novels seemingly written by rote?


Copy and paste your text into the window, and get a full report on word frequency, adverbs, passive voice and a fully marked up copy. Using similar programming to your word processor’s spell and grammar check, it confirms how rules-based language really is.


Pronounced ‘song lee’, this software engine from Japan can recognise the underlying structure of a song, tagging signals like the break, chorus, chords and melody. It raises the possibility of data from within songs rather than just genre, artists and title metadata because the computer will recognise pieces of and ‘understand’ music like we do.

Full client and publication list:

  • 3D Artist
  • APC
  • Auscam
  • Australian Creative
  • Australian Macworld
  • Australian Way (Qantas)
  • Big Issue
  • Black Velvet Seductions
  • Black+White
  • Bookseller & Publisher
  • Box Magazine
  • Brain World
  • Business News
  • Business NSW
  • Campaign Brief
  • Capture
  • Cleo
  • Cosmos
  • Cream
  • Curve
  • Daily Telegraph
  • Dark Horizons
  • Dazed and Confused
  • Desktop
  • DG
  • Digital Media
  • Disney Magazine
  • DNA Magazine
  • Empire
  • Empty Magazine
  • Famous Monsters of Filmland
  • Fast Thinking
  • FHM UK
  • Film Stories
  • Filmink
  • Follow Gentlemen
  • Geek Magazine
  • Good Reading
  • Good Weekend
  • GQ
  • How It Works
  • Hydrapinion
  • Inside Film
  • Loaded
  • M2 Magazine
  • Marie Claire Australia
  • Marketing
  • Maxim Australia
  • Men's Style
  • Metro
  • Moviehole
  • MSN
  • Nine To Five
  • Paranormal
  • PC Authority
  • PC Powerplay
  • PC Update
  • PC User
  • PC World
  • Penthouse
  • People
  • Pixelmag
  • Popular Science
  • Post Magazine
  • Ralph
  • Reader's Digest
  • ScienceNetwork WA
  • SciFiNow
  • Scoop
  • Scoop Traveller
  • Seaside Observer
  • SFX
  • Sydney Morning Herald
  • The Australian
  • The Retiree
  • The Sun Herald
  • The West Australian
  • TimeOut
  • Total Film
  • Video Camera
  • Video&Filmmaker
  • Writing Magazine
  • Xpress
  • Zoo