Can computers create?

Monday, March 27, 2017 - 6:01pm


Universal Pictures


Can computers think?

Alan Turing, the closest thing computer science nerds have to a god, first posed this question in 1950.

Today, I ask: Can computers create?

Before I dive into the second question I want to talk about the first one: Can computers think? Computer scientist Edsger W. Dijkstra countered this with another question: Can a submarine swim? And by extension, can an airplane fly? All three examples involve technology achieving a certain naturally defined endgame. Submarines swim, but not like fish. Airplanes fly, but not like birds. Resultantly, computers think, but not like humans.

We see submarines and planes as successful because they are able to navigate the seas and skies. To reach a similar bar of success, a computer that thinks should make decisions as rationally as human do. 

Computers can’t do this yet, but the entire field of Artificial Intelligence has emerged in pursuit of this goal. And in the past decade it has exploded.

Given that in a mere half century we have seen computers evolve from cluttering entire rooms to fitting into our pocket, making a machine that thinks seems like a realistic goal. Google Director of Engineering Ray Kurzweil, an incredibly accurate technology forecaster, said at the SXSW conference last month that this moment could be upon us come year 2029.

So if we wait a couple decades, there’s a good chance we’ll be able to definitively answer yes to the first question. And although this might seem menacing, I assure you there’s nothing inherently dangerous or scary about AI. Ex Machina is fiction. At its core, AI is a collection of algorithms and agents like Siri and Alexa are just a bunch of lines of code.

Now that we’ve theorized about the first question, we turn to the arguably more compelling question: Can computers create?

It’s one thing for a machine to drive itself across the country, but can a machine record the next 10/10 Pitchfork album? (Can anyone for that matter? I won’t launch into that debate right now though.) Or can a machine create the next Starry Night? For a computer to be innovative, its abilities must span far beyond mere thought.

The world has already seen some early attempts at creative AI agents, especially in the music field. Last week, a high school student debuted a machine-learning library that uses Kanye West lyrics to create its own ad-libs. Last year, Flow Machines, a project funded by the European Research Council, released two songs composed by a computer, though the production, mixing and lyrics are human-generated.

Despite the buzz and flashy headlines, the resulting compositions fall far short of the goal of creativity. The computer can rhyme, but the rhythm structure and wordplay is lacking. The Flow Machines tracks, developed in the style of The Beatles and Duke Ellington, sound like they came from their respective influencer’s discography. There was nothing new in the actual composition.

But just because these AI agents don’t meet the standard, doesn’t mean that none ever will. To reason about the future of this technology, we must consider what creativity is.

Steve Jobs said, “Creativity is just connecting things.” I think he’s pretty spot on. No musician exists without influences; no painter creates without predecessors. Artists look at what already exists — be it current events, society or other art — and rearrange select pieces in a way that has not been done before.

As we learn more about the neuroscience behind creativity, we will eventually pinpoint what makes a high-quality creative connection and develop technology that reflects this. It is likely that someday machines will be able to create like humans.

This doesn’t mean that artists will be obsolete, though. As I said before, there is something inherently different about creativity and thought. A self-driving car is successful if it drives, but an artificially intelligent artist is not successful if it merely creates something.

Artists are so much more than their art; they impact society and pop culture in ways that machines cannot. Kanye West, Lady Gaga, Frank Ocean — hell even Taylor Swift — all have reputations and societal impact that is infused into their music. You can’t listen to Kanye without thinking about his latest Twitter rant, or Swift without considering her latest breakup.

Machines don’t have emotions or personality. They could contribute industry-changing innovation, but the entire experience would fall short of a human artist’s.

Regardless, computers have the potential to compete with humans creatively. We shouldn’t see this as a threat though. Some of the best art results from competition. Considering that for years now the arts have struggled to be taken seriously by government and society, a creative revamp might be just what the industry needs. Although we might have to wait a while for it, creative AI could eventually spark a new artistic golden age.