News organizations are fighting for attention. They’re competing against technology companies that are much more adept at personalization. Netflix, Amazon, and Facebook have created addictive products by predicting users’ future behavior and honing in on their preferences. The more data they gather, the more they refine this process. If media organizations want to capture some of that attention, they have to improve how they personalize their audiences’ experiences. Artificial intelligence presents a way to offer hyperlocal, personalized, and niche stories — without putting more pressure on already overloaded reporters.

In June, I participated in a fascinating roundtable at Columbia University’s Tow Center for Digital Journalism. It was a gathering of technologists, journalists, academics, and legal scholars discussing the implications of artificial intelligence in the news industry. Many of the questions raised involved whether newsrooms would be hesitant to embrace AI: How can we encourage collaboration between journalists and technologists? How do we change the culture for people to want to do this?

Some reporters may worry about the impact of artificial intelligence on their jobs. They may balk at the idea of an algorithm being able to rapidly write thousands of news stories. But many of the use cases for AI would make reporters’ lives easier and their jobs more fulfilling. For example, the Associated Press worked with firm Automated Insights to increase twelvefold the number of corporate earnings stories it produces. It estimated that it freed up 20 percent of reporters’ time to work on more complex stories. Reuters also produces thousands of automated stories each day, in multiple languages, from corporate and government data.

Meanwhile, The Washington Post used its robot reporter, Heliograf, on Election Day to cover congressional and gubernatorial races. The technology also allows the Post to cover all D.C.-area high school football games every week. BuzzFeed was able to supplement its on-the-ground reporting from last year’s political conventions by deploying a bot that interacted with attendees. These cases show how journalists can benefit from automation helping to expand coverage while freeing reporters up to do higher-level work.

There is evidence that AI may improve the quality of journalism itself. “In the case of automated financial news coverage by AP, the error rate in the copy decreased even as the volume of the output increased more than tenfold,” said Francesco Marconi, who co-leads AP’s automation and artificial intelligence efforts.

Like John Keefe, I expect to see more major scoops next year by reporters using machine learning. The Atlanta Journal-Constitution was a Pulitzer Prize finalist for its yearlong investigation on sexual abuse by doctors. They analyzed more than 100,000 disciplinary documents and other records to find cases that may have involved doctors’ misconduct. They used machine learning to analyze the cases and assigned each case a probability rating, based on keywords, that it was related to sexual abuse.

Local news organizations are essential to their communities, but they struggle to cover them amid layoffs and shrinking newsroom resources. AI could be used to automate some types of local stories, such as crime reports and coverage of school and community board meetings. This would give residents access to more hyperlocal, personalized stories, while freeing up reporters to focus on other projects.

Most newsrooms simply do not have the resources to invest in artificial intelligence. That’s understandable. It’s time consuming, and the payoff may not be immediately apparent. Collaboration will be critical to advance how newsrooms use AI. That could mean a consortium of newsrooms working together on ambitious machine learning projects. Academic institutions have already led much of the research surrounding artificial intelligence, so we could also see more news organizations partnering with universities. However it happens, AI will become increasingly common in newsrooms until it’s seen as an essential part of the reporting process.