Artificial intelligence is already being used in newsrooms across the country to write straightforward news stories and unearth data trends. Does that mean that eventually, as machines become more intelligent, robots will replace reporters?

Not so, says Northeastern assistant professor John Wihbey, who is writing a book on the future of news in a networked world. But journalists are going to have to refine the concept of what is compelling news.

The stories that will rise to the top are the ones that only humans can unearth: stories that are analytical, critical, that synthesize the infinite amount of information available online.

“My hypothesis is that even as computers are able to do more of what we now consider news, capital-n ‘News’ will move up the value chain,” Wihbey said. “It will become more analytical, more critical, and potentially increasingly opinion-oriented.”

The rise of artificially intelligent technology and the sheer volume of information freely available online has forced a change in the role of journalists. They are no longer the “gatekeepers,” the only ones filtering information for the general public, Wihbey said. They’re becoming more akin to locksmiths.

Journalists “are the people who have the skills, the right approach, and the deft expertise to open a portal that otherwise nobody has a key for,” Wihbey said. It’s journalists who know where and how to look for data that’s not publicly available, who can be called upon for access to information the public is otherwise locked out of.

My hypothesis is that even as computers are able to do more of what we now consider news, capital-n ‘News’ will move up the value chain. It will become more analytical, more critical, and potentially increasingly opinion-oriented.

John Wihbey, Assistant Professor of Journalism

Artificial intelligence is already changing newsrooms. The Washington Post has used artificially intelligent technology to spit out short reports on the Rio Olympics, congressional and gubernatorial elections, and high school football games. The Los Angeles Times, The Atlanta Journal-Constitution, and Reuters are using it to parse through and synthesize huge amounts of raw data to look for trends.

In 2014, the LA Times published a massive investigation into police records, revealing that the Los Angeles Police Department had misclassified nearly 1,200 violent crimes as minor offenses. The paper’s reporting included the use of a machine learning algorithm that combed through eight years of crime data from reporters’ public records requests for indications that a crime had been misclassified. Reporters then manually checked the results.

“This is task reduction,” said Wihbey, who studies digital media, and whose book, The Social Fact: News and Knowledge in a Networked World, will be published in 2019. “It took a task that would’ve required thousands of hours of reporting time down to hundreds of hours of reporting time.”

At its core, artificial intelligence is “based on doing ridiculous amounts of multiplication problems very fast,” Wihbey said. Computer algorithms can be trained to calculate the probability that certain words will appear next to each other and use those probabilities to construct full sentences.

“As computers get better and faster at this sort of multiplication, things like automated writing become more possible,” Wihbey said.

Indeed, it’s hard to tell that humans don’t write the sports and financial markets wrap-ups compiled by machines for The Associated Press and The Washington Post.

“The Yorktown Patriots triumphed over the visiting Wilson Tigers in a close game on Thursday, 20-14,” begins one about a high school football game.

Though the writing styles among robots and humans can be similar, it’s the content of the stories that will differentiate them.

Back in the day, newspapers employed writers whose jobs were to recap stock market reports and weather forecasts. This information is still useful to readers, but it’s not necessarily considered “capital-n ‘News,’” Wihbey said.

“The kinds of automated stories we can imagine being produced will happen, but will start to become part of the information architecture that isn’t necessarily news,” Wihbey said.

That’s because humans are inherently “meaning-making creatures,” Wihbey said. “We want to co-create meaning, we want to connect around meaning.”

This sort of meaning-making is already happening. Many local, daily newspapers cover municipal government meetings as the paper of record in their communities.

Often, however, these meetings are available for viewing afterward, or televised live. Recapping them in writing doesn’t add much value for the residents and subscribers, and represent a bad investment of resources for newspapers competing for an audience.

“Already, the reporter has to think about what she can add to the conversations that’s genuinely valuable,” Wihbey said. Writing a valuable story about a publicly-televised meeting requires some new insight, some information that’s not readily available. It requires someone to unlock the more compelling stories hidden inside.

Wihbey said he has a hard time imagining that computers will ever satisfy a fundamental human need for meaning-making and storytelling—good news for human reporters. It is possible, he said, that the news industry will shrink considerably as algorithms become more and more capable of doing the more routine aspects of reporting. New jobs will likely emerge, however, as spin-off economies crop up. Journalists will need to become fluent in working with algorithms and harnessing their power to tell compelling stories.

“This is the story of technology and society,” Wihbey said. “Anything that can be automated, will be.”

But not everything can be.