Journalism Need Not Fear Artificial Intelligence

“I don’t think AI is ever going to replace journalists, and certainly not great journalists,” echoes Rubina Fillion, the director of audience engagement at The Intercept.

“What AI is good for is freeing up people who can spend less time focused on mundane tasks a lot of journalists end up doing,” Fillion continues. “It’s really a way to increase productivity and put more resources into reporting.”

AI is being used is many different ways in modern newsrooms, from generating story ideas from trends in social media or choosing a few images related to keywords in an article so a photo editor doesn’t have to sort through hundreds or even, thousands of shots.

One of its most common uses currently is automated text generation. Or, Broussard describes it: “Really, really, really sophisticated and amazing Mad Libs ... You have a template, and you fill in the blanks with words. That’s actually how these computer programs work, too.”

Think of sports, weather, or financial news that relays important numbers and figures. Many of these have already been automated.

“The news organizations that are using it most regularly are Bloomberg and the Associated Press,” Broussard says.

“What AI is good for is freeing up people who can spend less time focused on mundane tasks a lot of journalists end up doing.”

The AP’s use of automation has been so successful that according to a report from Harvard’s Nieman Journalism Lab, they’ve managed to produce 12 times as many earnings report stories while actually lowering their total error rate.

“A human writes a template, and then the structured data gets fed in. If you look at a lot of these stories together, they’re all pretty much the same,” Broussard says. “It’s great because it’s really boring to write the same story over and over again, but computers are really good at routine tasks that don’t change very much.”

While The Intercept prides itself on preserving the privacy of its readers, Fillion notes that other publications and news aggregators could use the sort of data routinely collected by many websites to offer readers incredibly tailored experiences, such as Amazon, Netflix, YouTube, Facebook, Instagram, and Twitter do already. “Adding that level of personalization in news reporting can help give us an advantage and make news more relevant to people,” she says. “You could create 5,000 or 10,000 versions of the same story by plugging in a person’s address ... It can be really hyperlocal content.”

Elsewhere in the newsroom, however, AI is turning journalists into cyborgs by augmenting the ability of investigative reporters and data journalists to find stories. Story discovery engines, like those that Broussard builds, are systems that enable reporters to efficiently parse through reams of data to uncover insights. “What the computer is really good at doing is processing the data and visualizing it in a way that helps you, the investigative reporter, identify what’s an anomaly and what’s not,” Broussard explains.

“It could help find out what the next Theranos or Enron is,” Fillion adds.

AI systems were instrumental to the International Consortium of Investigative Journalists’ Pulitzer Prize winning coverage of the 2016 Panama Papers scandal. Buzzfeed News used AI track the paths of spy planes over the United States last year. ProPublica used AI to figure out what individual U.S. lawmakers were most interested in based on what they talked about more than their colleagues.

The limits of how AI systems can be used seem to be limited only by available data and the imaginations of the journalists designing them. That’s part of the reason why Fillion felt it necessary to include women and women of color (including Broussard) when organizing the “AI and the Future of Journalism” panel for SXSW 2019’s Media & Journalism Track.

“It’s really important that you try to make sure that you have people from a variety of backgrounds working on this technology,” she says. “Because the worst thing that can happen is if AI, which is supposed to be the future, ends up reinforcing the power structures that already exist.”

At a time when newsrooms faced with dwindling revenues are being asked to do more with less, the tools offered by AI would seem like a godsend. But there’s less AI happening in journalism than you might think. The reason? Broussard says that though it comes down to labor issues, it’s really the inverse of robots pushing journalists out of their jobs.

“The journalism industry has had problem for many years of brain drain,” she explains. “I know a lot of journalists who have developed tech skills as an ordinary part of doing journalism, and then they discover that they can get paid a lot more in the tech industry, so they leave. It’s not part of the narrative but it should be.”

The AI and the Future of Journalism panel will be held on Sunday, March 10, as part of the Media & Journalism Convergence Track at the 2019 SXSW Conference.