Robot Reporters: Inside Automated Reporting

By Cameron Ellis, Special to the News Media Alliance

In an era when computers can fly planes, drive cars, and heat homes, why can’t they write the news?

That was an oversight that Joe Procopio, Chief Product Officer at Automated Insights, wanted addressed. He met CEO Robbie Allen, both of whom shared a belief that no one had taken a serious stab at that type of technology. Then a writer, Procopio and the rest of the team at Automated Insights saw a chance to reimagine the way natural language generation (commonly referred to as NLG) was used. NLG is, as Procopio describes it, “the process of generating narrative from, in our case, structured data.” As is often the case, Automated Insights tested this new structured data with sports journalism.

“[We] started by using NLG to create narrative from sports data,” he said. “We began by creating over 800 websites centered around professional and college football, baseball and basketball. Using the constantly updated data, we could generate new content for each one of those 800 sites every single day, up to five times a day.”

Since then, the company has expanded quite a bit. Today, Automated Insights automates fantasy football recaps, corporate earnings, financial and business reports, product descriptions, and even real estate market evaluations.  Procopio estimates that the company produced over 1 billion pieces of automated content in 2016. He’s excited about the work that Automated Insights is doing.

“The biggest advantage to automated reporting has to be streamlining some of the more mundane data-driven tasks in journalism,” he said. “By automating content that is largely statistical, a company has the ability to produce reports, written to the highest standards, without the journalist having to be a data scientist. This allows journalists to give deeper thought on reporting that is more valuable to the story.”

“There are three additional very distinct advantages as a whole,” Procopio added. “The first being reach. We give organizations the ability to report on the long tail, entities and events that they don’t have the human resources to cover. Secondly, depth. We allow people to discover trends and insights within data that would require a human to spend a lot of time, often with highly specialized software. The third advantage is speed. We can create personalized content in a matter of mere seconds with a ton of volume. We always say, ‘We don’t write one article that’s intended to be read by a million people, we write millions of articles, each one intended for a single person.’”

The company continued to show its dedication to reimagining journalism with a sizeable partnership with the Associated Press. It’s a savvy pairing – especially, according to Procopio, when it comes to business reporting. Prior to the partnership, each quarter the AP would assign writers the job of going through corporate quarterly earnings reports and summarizing the data. Even a staff with the size and resources of the AP couldn’t generate more than 400 reports a quarter. Now, with Automated Insights technology, they can generate over 4,500 reports per quarter – a 1,125 percent increase.

For all the pros of news automation, Procopio says there are also a few cons.

“[They] are only as good as the data.” Essentially, restructured or inaccurate data can lead to a bevy of complicated issues.

While some – especially those still attached to traditional journalism – see automated reporting as a threat to the field, Procopio envisions a world where the two work hand-in-hand.

“I believe it will become the most prevalent tool in the journalist’s toolkit,” he said. “I also see automated content finally delivering on the promise of the internet being ‘push versus pull.’”

When he says ‘push versus pull,’ Procopio is talking about putting together the content provided in the right context. He imagines a scenario so seamless that “you can take action immediately, as opposed to having to search for what you need to know.” As for how he would answer those who continue to be skeptical that they’re losing their column inches to a Macbook Pro?

“Journalism may be contracting, but it’s also evolving,” he said. “A lot of the grunt work of journalism can now be done by machines. This means that the best journalists will continue doing what they do, and we can help them do it better.“

×

Join the News/Media Alliance for a Lunch Briefing on 4/11 on 'Respecting Creative Industries in the Age of Generative AI' - Learn more