media update’s Adam Wakefield unpacks how newsrooms around the world are using machine learning and AI to make their organisations better, faster and stronger.

How the LA Times used machine learning to interrogate statistics

In a report found on Medium.com, Freia Nahser, innovation reporter and editor of Global Editors Network, explains how the Los Angeles Times used machine learning algorithms to show how the city’s police department misclassified 14 000 serious assaults as minor offences between 2005 and 2012.

This statistical change effectively lowered the city’s crime rate, when the reality was somewhat different. Nahser says, “The Los Angeles Times used an algorithm that parsed crime data from a previous Times investigation in order to learn the keywords that identify assaults as either serious or minor.”

“The trained algorithm was then let loose on a random sample of almost 2 400 minor crimes that took place between 2005 and 2012 to find which of these assaults were misclassified.”

The results were then manually checked, as the algorithm carried a 24% error rate. The newspaper then altered the estimated tally of misclassified crimes based on the error rate, with the final results concluding that violent crime in Los Angeles was 7% higher, and serious assaults 16% higher, than reported by the police department.

Reuters uses machine learning to validate real news on Twitter

Over a two-year period, news agency Reuters developed a tool called Reuters News Tracer, which helps journalists spot and validate real news in real time on Twitter. The tool receives alerts that allow the wire service to contact eyewitnesses to “see what’s happening around the world”, the agency explains on its website.

“Crucially, this is providing more time to do value-added reporting work.”

Through leveraging cognitive computing and machine learning, Reuters News Tracer finds insights from vast amounts of social media data, combining algorithms that “merge artificial intelligence with the human intelligence of Reuters journalists”.

The tool runs machine-learning algorithms on a percentage of Twitter’s daily tweets, numbering in the hundreds of millions, to find breaking news. It does this by looking for clusters of tweets discussing the same event. A newsworthiness rating is then attached to an event, with the tool then reverse engineering how a journalist would verify whether a piece of information is true or not.

The agency’s journalists then independently verify the information through their own channels and reporting – before publishing. 

The Washington Post’s robot reporter

In September 2016, The Washington Post launched its own in-house AI technology called Heliograf. According to Lucia Moses, a senior editor at Digiday, Heliograf was designed to produce approximately 300 short reports and alerts for the Rio Olympics in 2016.

“Since then, it’s used Heliograf to cover congressional and gubernatorial races on Election Day and DC-area high school football games”, producing tweets at the same time.

Jeremy Gilbert, director of strategic initiatives at The Washington Post, told Moses that Heliograf produced around 850 articles for the paper in its first year of operation, including 500 articles focusing on the 2016 elections – generating more than half a million clicks in total. While not a lot of clicks in the broader scheme of things, the majority of these stories were not assigned to the staff.

“For the 2012 election, for example, the Post did just 15% of what it generated in 2016,” Moses says.

“Media outlets using AI say it’s meant to enable journalists to do more high-value work, not take their jobs. The AP [Associated Press] estimated that it’s freed up 20% of reporters’ time spent covering corporate earnings and that AI is also moving the needle on accuracy.”

Moses says The Washington Post is looking into how it can use Heliograf to help its journalists do substantive reporting.

Want to stay up to date with the latest media news? Subscribe to our newsletter.
Like other organisations in the news space, The Wall Street Journal is also using machine learning to advance itself as a news organisation, specifically its paywall. For more information, read our article, How The Wall Street Journal is using a smart paywall to find subscribers.