Earlier this year the Los Angeles Times published a story about an earthquake. Nothing too unusual in that, you might think – California experiences thousands of them a year. But this story was a little different: it was written by a robot.
The article looked like this:
‘A shallow magnitude 4.7 earthquake was reported Monday morning five miles from Westwood, California, according to the U.S. Geological Survey. The temblor occurred at 6:25 a.m. Pacific time at a depth of 5.0 miles.
According to the USGS, the epicenter was six miles from Beverly Hills, California, seven miles from Universal City, California, seven miles from Santa Monica, California and 348 miles from Sacramento, California. In the past ten days, there have been no earthquakes magnitude 3.0 and greater centered nearby.
This information comes from the USGS Earthquake Notification Service and this post was created by an algorithm written by the author.’
What do you think? A little dry, perhaps, a little prosaic. It does the job for sure – telling you there was an earthquake measuring 4.7, five miles out of Westwood.
If we were being picky, we might query the use of the word ‘temblor’. It’s another name for ‘earthquake’, but how often do you see it in news reporting? And then there’s that final bit…
Rise of the machines?
‘Robo-journalism’ is becoming more and more prevalent in newsrooms and among publishers all over the world.
Typically, it works by inserting data into a ready-made template. So the template for the above earthquake story might read something like this:
‘A [ ] earthquake was reported [ ] miles from [ ] according to the U.S. Geological Survey. The temblor occurred at [ ] Pacific time at a depth of [ ].’
And so on – the data is simply mixed in afterwards. That’s why robo stories are, at core, always based on data – like crime statistics, sports scores, earthquake magnitudes and stock prices.
Algorithms are appealing to publishers – they’re a way to get a lot of information out there, fast, accurately and, of course, cheaply. But can software really write news? Or at least, news beyond a stock market fluctuation? Could a machine write something that instils emotion, conveys experience? And should journalists be worried about their jobs?
The next generation
This week the Associated Press announced it is to team-up with a firm called Automated Insights in a move that will see robots write some AP business stories from July.
The shift will be massive: instead of publishing 300 earnings reports (each 150 to 300 words) every quarter, the company will provide as many as 4,400 – in roughly the same time it would take AP reporters to write 300.
“We believe technological automation will be a part of many businesses, including those in media,” said Lou Ferrara, AP vice president and managing editor.
So goodbye writing jobs for humans? The argument for robo-writing posits that instead of job cuts, it will actually revolutionise media, creating a much richer stream of content where human journalists focus on producing valuable, meaningful stories – leaving the bots to bash out the data.
“This is about using technology to free journalists to do more journalism and less data processing, not about eliminating jobs,” Ferrara said.
“Instead our journalists will focus on reporting and writing stories about what the numbers mean and what gets said in earnings calls on the day of the release, identifying trends and finding exclusive stories we can publish at the time of the earnings reports.”
Here’s another example of robo copy:
‘Friona fell 10-8 to Boys Ranch in five innings on Monday at Friona despite racking up seven hits and eight runs. Friona was led by a flawless day at the dish by Hunter Sundre, who went 2-2 against Boys Ranch pitching. Sundre singled in the third inning and tripled in the fourth inning … Friona piled up the steals, swiping eight bags in all …’
As Wired reports, this content – a write-up of a US Little League baseball game – comes from Narrative Science, a Chicago-based “intelligence reporting” firm that ‘trains’ computers to write news.
Not bad. Certainly an improvement on the earthquake piece. We’ve even got terms such as ‘racking up’ and ‘flawless day at the dish’. Narrative Science created the article by using pitch-side game data that parents entered into an iPhone app called GameChanger.
In February this year Christer Clerwall of Karlstad University in Sweden published research looking at how readers perceive robo-generated news content versus news content written by journalists.
The study saw readers given a mix of different articles written by humans and computers. The participants were then asked to rate each article against several criteria including quality, credibility and objectivity.
The copy written by a human journalist was found to be coherent, well-written and pleasant to read, while the computer-generated content was seen as descriptive and boring. But it was also found to be trustworthy and, overall, readers struggled to differentiate between the two.
The question is whether it could evolve. Sure, an algorithm could be improved – it could get more sophisticated, more able to write fluidly. But could robo-writing ever extend beyond the realm of data? Would we one day see an algorithm write a long-form piece of content?
Rapid technological change is altering the way we create and consume media. But will software-generated content ever truly stand shoulder-to-shoulder with human journalists?
This information comes from TranslateMedia and this post was created by a human being.