By Chris Adams
At The Washington Post, hundreds of smart journalists work essentially around the clock to bring readers news and insights.
Those journalists are generally the best in the business. But how will they fare against a computer?
Joey Marburger, director of product at the Post, led National Press Foundation fellows through a discussion of the use of artificial intelligence systems in the creation and curation of news. Marburger (bio, Twitter) is one of the innovators behind the Post’s online, mobile and video offerings in recent years.
Marburger described Heliograf, the Post’s intelligent, automated storytelling agent. It can generate stories from real-time data sources and personalize them for readers.
It’s not replacing the smart reporting and analysis of human journalists. But it can handle the simplest of stories. Right now, that means book best-seller lists, high school sports and biographical sports “cards” for National Football League players.
“We can’t cover every single high school sport – but we’d love to,” Marburger said. This allows the paper to get closer to that goal.
In the elections of 2018 (Marburger spoke a week before the midterms), the Post plans to use automated systems to update races, the potential balance of power in Congress and nitty-gritty details key insights political junkies love.
“It can’t really tell stories like a journalist can,” Marburger said of Heliograf. “But it can write briefs.” He compared it to the children’s word game Mad Libs, where the blanks in a pre-formatted story are filled in.
“Heliograf is like a higher order of Mad Libs,” he said.
Marburger also talked about tools that combined photos and text to produce simple videos of news stories. Those aren’t yet part of the Post’s broad news products, but they are being tested. And that product – HelioVideo – shows the potential impact on journalists and their jobs.
To produce one of those simple videos takes the Post’s best video producers about 30 minutes, Marburger said. It takes HelioVideo five minutes.