Robo-journalism isn’t ruining the media, it’s helping journalists. Here’s how.By on November 10, 2017
Much has been made of robo-journalism and its implications for media businesses and journalists. Some call it the death knell of journalism, others say it will usher in a new era of reporting capabilities. The real answer is probably somewhere in between the two.
First, a quick definition of robo-journalism, at least for our purposes here. No, it doesn’t mean sending an android to cover a baseball game or a press conference. Though that would be cool. Robo-journalism is the process of having a computer write partial or complete news stories without human intervention.
The AP Example
Associated Press is probably the best example of a fully implemented newsroom AI solution. The news wire uses a program called “Wordsmith”. Wordsmith turns nominal and numeric data such as rosters, game statistics, and scores into templated stories for online news. According to The Verge, the AP tried covering minor league games more than a decade ago, but couldn’t cover the full schedule of teams and games.
Now, using a data feed from Major League Baseball, the Wordsmith platform can cover 142 teams in 13 minor leagues without sending a reporter. Wordsmith allows AP to provide coverage it would otherwise not be able to because of staffing shortages.
AP is also using Wordsmith to write up financial earnings reports, publishing 3,000 stories per quarter and growing. Before the implementation, the news organization was only able to cover about 300 companies in the same amount of time. AP is also careful to note that when the program began, every automated story had errors logged and sent back for tweaks. The Verge reports the automated system now logs fewer errors than the human-produced stories from years past.
So what does it mean?
It is important to note that the venerable news wire has implemented Wordsmith without eliminating a single job. In fact, AP has hired its first “automation editor” to oversee the process and discover other efficiencies within the organization. And we’re not talking about efficiencies that get reporters laid off. These efficiences, as automation editor Justin Myers says, give reporters time to be, well, reporters: “In a lot of cases [they’re] slogging through whatever processes we’ve built up over the years, rather than focusing on doing journalism.” “Let’s have a computer do what a computer’s good at, and let’s have a human do what a human’s good at,” he says.
Having a computer write up earnings reports also removes a key burden from financial reporters: writing up earnings reports. USA Today business reporter Roger Yu calls them “the quarterly bane of the existence of many business reporters.” AP’s managing editor, Lou Ferrara, also points out that the news wire continues to produce staff-written earnings stories for high-interest companies, such as Google and Apple.
Training is key in robo-journalism
Most importantly, we should note that Associated Press didn’t simply install Wordsmith and click “Generate News”. AP spent a year testing the software to make sure its minor league game reports made sense. Human reporters and editors worked for months with Wordsmith to teach it how to report on earnings: What milestones are important? Did the company exceed or fall short of analyst expectations? A computer doesn’t simply know these things. AI training, feedback and corrections, and constant updates to the algorithms are all vitally important.
So, at least until the androids show up at a press conference, AI is not killing journalism. It’s letting reporters dig deeper into stories that matter more, and leave the minor league baseball and small-cap earnings reports to the robots.