Robots may be writing more stories – but journalists are skeptical that they can move hearts and minds the way humans can.

 

Today, the passion of the journalist is being put to the test. Nearly half of current jobs could be computerized by 2034, according to a 2013Oxford University study– and creative work like journalism won’t be spared.

Developed at Northwestern University, the artificial intelligence software StatsMonkey generates sports reports by transforming data into articles, ostensibly without human help. These natural language technologies open the door to so-called “robot-journalism.”

More and more major media outlets, among them the Associated Press, the New York Times, Forbes, the Los Angeles Times and the Washington Post, have adopted automated story-producing software.

“It just might be the worst thing I’ve heard,” said journalist Lenn Robbins, who worked for more than 30 years covering professional and college sports for a range of publications, including The (Bergen County, NJ) Record and the New York Post.

“Journalism is about getting accurate information through research and interviews,” he argued. “It’s about building sources and trust.”

Nor are journalism students happy about the prospect of competing with robots.

“[Automation] is more likely to happen in the United States, specifically because we have for-profit media,” said Brittany Gibson, a journalism student at Rutgers University. Gibson’s dream is to work as a foreign correspondent. But she sees that it is getting harder to get any journalism job, especially in the United States.

An article generated by a robot.
An article generated by a robot.

It’s true that automated technology has made journalism workflows more efficient. The Associated Press has used an automatic news writing software called Wordsmith, marketed by Automated Insights, since 2014. The AP reports that it now produces 4,400 quarterly earnings stories each quarter–a twelve-fold increase over when only human journalists produced such reports.

Automated journalism also allows media outlets to reach small audiences, by inexpensively producing articles likely to interest only a few readers. For example, the Washington Post uses an automated software called Heliograf to cover many high school football games in the Washington, D.C., area.

Robot journalism also allows news organizations to quickly and rapidly generate articles from large databases, such as those linked to financial reports and election returns.

“That’s something everyone wants and needs to do in order to survive,”  Journalism Education Association President Sarah Nichols acknowledged.

“Data journalism, if anything, is experiencing a bit of a renaissance,” said Rutgers University journalism professor Rachel Kremen, who teaches courses in data journalism. “I hope people stay engaged with thoughtful data stories, especially those that fully explain the data and put it in context.”

 

“Think pieces and investigative stories are essential to a functional democracy. But they are also a bit like broccoli: they might be good for you, but you don’t often see people clamoring for them.” –Journalism professor Rachel Kremen 

Robbins, who teaches sports journalism at Rutgers University and New York University, has never considered sports reporting endangered automated article writing.

“Fans want to have a connection with their athletes,” he said. “They want to hear what they have to say, and to learn about what makes them tick. “The day automated journalism can do that — bring an athlete to tears, or answer a question with depth and insight — is the day I move to New Hampshire and open a dog rescue shelter.”

Rutgers University journalism student Josette Rogers likewise doubts automation will threaten the kind of journalism humans produce.

“Society will always need people to write about current events that have the potential to spark debate, and automated technologies will not have that kind of power,” she argued. “I would like to learn how to evoke emotions through my words in the various pieces that I will be writing in my career.”

The L.A. Times tweets to apologize for an erroneous earthquake article generated by the Quakebot algorithm.

Machines have also made eye-popping mistakes: The Los Angeles Times, via an algorithm called Quakebot, in 2017 tweeted an erroneous earthquake report, based on data from 1925.

Still, as news organizations see benefits, the pace of technological innovation will increase. And robot journalism is not going away.

“It’s not a matter of competing,” Kremen said.“My concern is about who will write and control the software.”

As long as robot journalism continues to grow, people in media outlets need to learn to control those technologies appropriately.

“Journalism schools should help prepare students with an understanding of how automation services work, so that journalists can help ensure automated services operate ethically and according to sound journalism principles,” said Jeremy Caplan, director of education at CUNY’s Craig Newmark Graduate School of Journalism. The school offers courses that help students develop coding and design skills, which can be used to refine or build on automation services.

Nichols is concerned about the ethical aspects of how, when or to what extent media outlets need to notify readers when content is automatically produced. It is true that people are more skeptical about the source, because of the rise of fake news. Readers do care about accuracy and transparency of stories. At the same time, readers also need to be educated in the ways auto-generated content differs from traditional journalism.

“The day automated journalism can bring an athlete to tears, or answer a question with depth and insight, is the day I move to New Hampshire and open a dog rescue shelter.”—Journalism professor Lenn Robbins  

“The challenge is whether or not people will pay for the other stuff, that humans are better at writing and researching,” Kremen said. “The trend seems to be for listicles and fluff pieces.” She wonders whether readers still have a big appetite for investigative stories.

“Think pieces and investigative stories are essential to a functional democracy. But they are also a bit like broccoli: they might be good for you, but you don’t often see people clamoring for them,” Kremen said.

Robot journalists generate articles many times faster than human journalists do. They analyze data more accurately than humans. So journalists can’t compete in those areas – they instead need to focus on cultivating “good stories,” several of those interviewed for this article agreed.

“What human journalists bring to the table is empathy and curiosity, and I’d argue those are the most important traits in finding and telling good stories,” Nichols said.

Even though robots can assemble data, they can’t write about whether a team “almost” won, or the frustrations players experienced.

“What makes good sports stories,” Gibson said, “is you can capture the ethos of what it means to lose, or what it means to win.”