We live in an era of self-driving cars and light-up hoverboards. It shouldn’t come as a surprise that robots are starting to replace professions that were once viewed as invaluable—financial advisers, surgeons and reporters all have automated equivalents.
This doesn’t mean that all journalists are going to be replaced by typing WALL-E replicas, but it is a technology that warrants caution.
“Robo-journalism” can generate thousands of stories in the time it takes a reporter to file one (over 1,000 articles per second is the standard). The Associated Press has been using a word-generating platform called Wordsmith for analysis-based writing since 2014, and Yahoo also uses the software for recaps and reviews of fantasy sport.
A robot-generated story is born in stages. First, data is inputted into the system. This information is comprehensive and comes in large quantities, but it’s limiting in that there’s no room for analysis beyond a refined scope. An algorithm determines what’s newsworthy by detecting changes in value and minimums and maximums (for example, changes in stock prices for business stories).
Next, the bots find an angle. This comes from a pre-authored library of story patterns based on events and circumstances. A way to approach a generic sports story, for example, might be “heroic individual performance,” “strong team effort,” or “came out of a slump,” wrote computational journalist Nicholas Diakopoulos in a blog post on the anatomy of a robot journalist.
From there, it’s a game of connect the dots: specific story points in the generic copy are matched to individual pieces of data, like the names of the players and the score. To incorporate personal details and other factual context, additional information is pulled from internet databases.
In 2014, Business Insider published an article called, “If you don’t think robots can replace journalists, check out this article written by a computer.” It followed the release of Quill, an artificial intelligence product created by Narrative Science. Scientist and co-founder Kris Hammond told Business Insider the product could “turn boring data and statistics into highly readable stories with a beginning, middle, and end.”
An expert from sample report about a baseball team called the Manalapan Braves Red, generated by the software, reads:
Cole Benner did all he could to give Hamilton A’s-Forcini a boost, but it wasn’t enough to get past the Manalapan Braves Red, as Hamilton A’s-Forcini lost 10-5 in six innings at Pecci two on Saturday.
Tech Times reported that AP produced around 4,300 stories per quarter in 2015 using Wordsmith — 14 times the amount of content than previous years. In an already fragile job market, that’s cause for concern. What happens when journalists become replaceable?
Martin Ford, a leading commentator on the potential impact job automation will bring to the economy, told TheMediaOnline that there will be no “slaughter of all the journalists.” A BBC infographic based on an Oxford University and Deloitte study about robots taking over jobs ranked automation for journalists and editors as “quite unlikely,” with an 8 percent risk factor.
But even if jobs aren’t being eliminated, the technology could create fewer entry-level positions, according to Ford. “If you want to have a career in journalism and you’re just graduating … your first assignment may be one of those routine formulaic things,” he said. “That’s been the way that journalists have learned the ropes from the beginning.”
Robo-journalism is still in its infancy. But as technology improves, it’s possible that news organizations will eventually start experimenting with automatic newswriting in other areas. Is that what we want? A good journalist can accurately analyze information, but a better journalist asks meaningful questions and thinks critically—things a machine will never be able to replicate.
About the author
Nicole Schmidt is the online handling editor of the Spring 2016 issue of the Ryerson Review of Journalism