Are robots coming for your job?
A computer that can write sport reports, a program that creates billboards and banner ads. Are machines muscling in on your job or are they just trying to free you up to do more of the good stuff? In a feature that first appeared in Encore, Matt Smith investigates.
So here’s a short news story from Forbes.com: “Analysts have become increasingly bullish on Apple (AAPL) in the month leading up to the company’s second quarter earnings announcement scheduled for Tuesday, April 24, 2012. The consensus earnings per share estimate has moved up from US$9.60 a share to the current expectation of earnings of US$9.86 a share.”
The structure might be somewhat formulaic, but it gets the job done and has all the numbers you’d expect in an article about stock market changes. Unlike most things you read, it isn’t written by a soft mushy human with its twiddley fragile fingers. It’s written by a computer algorithm which sorts through data, looks at patterns, and presents a story with an acceptable amount of prose. \
And it didn’t even need to use a keyboard. The engine is called Quill, an advanced writing algorithm developed by Narrative Science, an American software company. Far be it for this precursor to Skynet – the artificial intelligence unit that featured in the Arnold Schwarzenegger film Terminator – to just be happy churning out finance stories, it’s also muscling in on reporting sport scores as well. Given adequate data, it can produce fast copy, regardless of the hour, the pay, or the caffeine content.
As a journalist it has its limitations. It still relies on what Narrative Science calls ‘meta-writers’, people to format templates for Quill to input data to. It’s output lacks linguistic flair, and falls squarely into the ‘adequate’ category.
But far from being intimidated by the prospect of a journalistic replacement that will literally work on the smell of an oil rag, Marcus Strom from the Media, Entertainment and Arts Alliance, the journalist’s union, sees Artifical Intelligence (AI) as a useful tool. “At the moment AI is expensive and difficult to operate,” he says. “In 10 to 20 years time that might be different. If it’s used correctly it can free up journalists for more in-depth, investigative journalism that can’t be done by an algorithm.”
Strom also points to the changing business models of newspapers as a reason robots with their so far basic skill sets are unlikely to put real journos out of a job. “People are going to need more than stories with numbers to get them through a paywall, and if you’re reading a story about a sports game, you want more than just the scores. That takes human interaction and journalists, not computers,” he says.
A machine’s ability to replace journalists might not be quite there yet, but technology has got a firm grasp on automating processes. While media agencies once purchased the bulk of their inventory from sales teams, advertisers are now increasingly being automatically matched with publishers through exchanges.
Online ads are already matched by automated media buying, and Danny Bass, chief investment and knowledge officer at GroupM, says it isn’t going to stop there. Outdoor and radio ad purchases are already automated in the US, and are about to be in Australia. The technology to automate buying for television will be in place within the year.
“There’s been an explosion in programming digital media, and that light is now starting to shine on traditional media,” Bass says. “Clients really like the results, and the reality is, there’s no way the sales team would be able to cope with the growing volume without automation.” Like Strom, Bass says that far from the technology being a threat to humans, it presents an opportunity. “Automation has freed up planners in some areas, which is good, because there’s only going to be more to deal with in the future. We’re booking more spots than ever with digital channels. So now it’s all about planning for the next five years,” he says.
So robots can write basic copy and match media buyers to publishers, this much we have established but they aren’t about to decommission anyone with a creative-based job. Or are they?
It turns out that there are a couple of robots who could be making creatives shift uncomfortably in their seats. The design department should be most fearful of AARON, a painting program developed by artist Harold Cohen. Cohen has been tinkering with it for decades and believes it will continue to paint in his style long after he dies. Another program with its eye on the art director’s office is The Painting Fool. The Painting Fool can pull news stories from Google and come up with a way to represent them artistically and programmer Simon Colton likens what it does to the way humans are inspired.
When asked if the painting fool could find a home in an ad agency, Colton cryptically answers: “Your questions are very interesting. In fact, I have a big new project starting in October to build the ‘WhatIf Machine’ which will undertake fictional ideation, and hopefully be of much use to the advertising industry.”
These programs could well lead to something bigger and there are others getting in on the game. An algorithm called CAI made headlines three years ago for its ability to create passable billboards and online banner ads based on a few inputs such as a product and a mood the user wants to illicit.
“At the moment these programs have been developed purely for research,” says Jon McCormack, electronic media artist and associate professor of computer science at Melbourne’s Monash University. “The programs are all about being fine artists, not commercial artists. The problem is, it’s just so far away from a product that can take a few words and come up with a picture that can be used in an editorial sense.” McCormack concedes that there are likely to be others tinkering away in labs all over the world to produce programs that could impact our industry but says creatives aren’t likely to be out of a job anytime soon. He says: “I think it will be possible in the next five to 10 years for computers to be creative, based on the current technology. Maybe one day. The short answer is it’s not there yet.”
This story first appeared in the weekly edition of Encore available for iPad and Android tablets. Visit encore.com.au for a preview of the app or click below to download.