Earlier this week, the New York Times had a story on its site about income mobility. It’s called the Best and Worst Places to Grow Up. The Times’ story was based on a 2014 Harvard study.
Andrew DeVigal, the chair of journalism innovation and civic engagement at the University of Oregon’s School of Journalism, said when he first stumbled on the Best and Worst Places to Grow Up, he was kind of surprised the Times wrote the story from Oregon.
“Like, ‘Oh, this is great. The New York Times is actually writing a story about Multnomah County.’ Until I started reading into it, and I was like ‘Oh, this was actually crafted for me,” DeVigal said.
The story is crafted for DeVigal. It’s also crafted for the reader living in San Diego, or El Paso, or Chicago or anywhere in the United States. DeVigal says that’s because the Times site recognizes the IP address of the user. Then the Times had a program that took the data from that Harvard study and wrote a story about that location.
“It was actually nicely, narratively written. It was customized. And it was sprinkled in the uniqueness of the article based on where I was, where my IP address was. But at the same time, every third, fourth, sixth paragraph was in context to the larger, national issue. And only a few of the graphs were customized to me. You know, it was really smart,” DeVigal said.
Stories powered by computer programs are just going to get smarter. While the Times utilized human writers for that particular story, other outlets, including the Associated Press, Forbes, The Los Angeles Times and others, are utilizing algorithmic programs that write stories autonomously without any human input.
Here’s how it works: data is fed into the computer program where it’s sorted, arranged, and then written as a story. And most of these stories are for data heavy subjects, like corporate filings, or sports summaries.
James Kotecki is a spokesman at North Carolina based Automated Insights, which makes those computer programs. “The colloquial way of describing what we do is that we make robot writers. But there are no actual hulking automatons in our office typing out prose, it is a software, we’re using software to do this,” Kotecki said.
Chances are good that you’ve read one of the company’s computer-generated stories. The Associated Press uses automated insights. Before the news organization began publishing robo-writers, it published about 300 corporate earnings report per quarter. Now, it’s publishing more than 3,000.
So does the company hate journalists and is trying to ensure more job losses in the field?
“We don’t, we do not hate journalists. In fact, journalists love us!,” Kotecki said with a laugh. “They were the most optimistic and excited about it because they said, ‘This is the kind of work we didn’t like doing in the first place.’
“So typically what we automate are things that, A, never could’ve existed in the first place without automation, so something that humans were never doing in the first place. Or, B, something that humans didn’t enjoy doing and what we’re doing is freeing them up to do more interesting work,” Kotecki said.
Those stories nobody was covering are things like weekly summaries for fantasy sports leagues. Does anyone disagree with this? We asked the The Poynter Institute, a journalism college and think tank. Poynter says it’s good with robo-writers. So we call other journalism experts. Time and time again, we get the same response: Robo-writers are fine. Especially because there is such a crushing demand for online, on demand content – any content. They all see robo-writers as helping to relieve that pressure.
Even at the Portland Mercury, a publication with a unique voice and tone, News Editor Dirk VanderHart said he’s not worried about robo-writers.
“The reason that I’m skeptical is that we really do value the ability to be smartasses and be conversational here and level with the reader and be personal. And that’s not something I’ve seen in an algorithm. I think what people are finding as journalism plummets is that there is a value in place for the really more meaty stuff and some of the more ‘let’s fill our website stuff’ is falling by the way side. At least that’s my opinion,” VanderHart said.
But what if these robo-writers did meaty stories? What if they were able to quote, or offer context? What if they added analysis like humans? Enter Paris based company Yseop.
Arden Manning, spokesman for Yseop says that’s exactly what their software does. Manning said they are working on software that explains motivation and context in computer generated stories. And he said the public and journalists shouldn’t fear robo-writers.
“A hundred years ago, an accountant had to do all of the budgeting on a ledger, a piece of paper. He used a pen to total up all those numbers. And then the calculator came along. Still had plenty of accountants, they just were much more productive. Then, fast forward a few years, Microsoft Excel came along. Still have the same amount of accounts, they just have a lot more clients and their expertise is available to a lot more people,” Manning said.
Still, even if we accept these algorithm writers as an improvement to journalism, there are other questions that couldn’t be answered. Like what happens if a hacker gets into a system, and a story is published saying an important world leader was killed? Or what if politically minded programmers skew the algorithm to write articles in line with their perspectives? And what does it mean that we’ve created a machine designed to tell our story as humans?
Andrew DeVigal from the UO journalism school said this development could spur journalists and the public to look at story telling as an art.
“In some ways I think it is the art form that really encourages us, and engages with us, in a much more emotional way. I actually think there are opportunities for us to really think of it as a tool, more than a threat,” DeVigal said.
And that’s thing about technology, it’s often seen as either a tool or a threat, but author Margaret Atwood reminds us they could be both. Last December, Atwood wrote in the New York Times, “Every technology we develop is an extension of one of our own senses or capabilities. The spear and the arrow extended the arm, the telescope extended the eye.”
She went on to write, “Every technology we’ve ever made has also altered the way we live.”
And so we’re left with the question, how will our lives change, should computer programs tell our stories?
Editors note: Special thanks to OPB’s Steven Kray, who helped produce this story.