“Technology is enabling us to do things that we couldn’t do before,” he said. “Our expectation is to free reporters’ time for them to cover their beats.”
The AP expects to increase their number of earning reports from 300 manually-written stories to 4,400 automated pieces about companies across the United States every quarter. The company’s stated goal is to make reporters’ jobs easier. When asked whether the algorithm could also help save the AP labor costs, Lou Ferrara referred to the company’s press release.
“No. This is about using technology to free journalists to do more journalism and less data processing, not about eliminating jobs. In fact, most of the staff has been receptive to the effort and involved for the past few months of discussion,” the AP’s blog post said.
The company Automated Insights has been building algorithms like this one for years, not only for news organizations but also for companies like Samsung, Microsoft and GreatCall. Scott Frederick, chief operating officer of Automated Insights, said that the goal is to provide personalized, engaging narratives while reducing reporters’ workload.
“We let the data tell its own story. Our added value for journalism is focusing on quantitative aspects of the news: the what, where and when,” said Frederick, “so that we can free reporters to work on the how and why.”
Some financial reporters welcomed the tool for alleviating their work with numbers. Other experts expressed concerns, though, that the quality of financial reporting could suffer.
“We still have to wait and see what happens with other players like Reuters, Bloomberg and Dow Jones,” said Greg David, director of the City University of New York’s business reporting program. “One aspect that concerns me is, if reporters are no longer monitoring the earning reports, that they might miss good stories on their beat.”
But some reporters who work on financial earnings praised the platform. “From the perspective of a journalist, I must say that this is pretty exciting,” wrote Jordan Crook in Techcrunch. “Earnings season is a stressful time for all media publications.”
Kevin Roose wrote for New York Magazine that “the stories that today’s robots can write are, frankly the kinds of stories that humans hate writing anyway.” He recalled his painful experience working on earning reports at the New York Times and said he is enthusiastic about how algorithms can process information and deal with tons of data that humans have a hard time handling.
The algorithm, in Automated Insights’ words, is “the next generation presentation layer for big data.” One of the challenges of reporters working with long data sets, Frederick said, is that not only do reporters get exhausted, but audiences also can’t manage information overload.
“People don’t like browsing dashboards they need to interpret,” Frederick said, “They would like to see pictures. They would like to be told stories.”
Although the AP and other media outlets have been working with automated feeds of data for a while now, this is the first time that an algorithm is using raw numbers to produce written pieces rather than just charts. With a label below that says “Generated automatically by Automated Insight” the AP makes clear which stories are created by robots.
Since every news outlet has its own style, this particular algorithm was trained and educated on the AP writing style, complying with the same rules given to reporters. This way, the robot digests structured data, coming up with an output that should be more consistent with the publication’s standards by simulating tone and style.
Can you tell which phrases were written by a human, and which ones by a robot? Take our quiz and let’s see if you can tell the difference.