The Washington Post employs more than 700 journalists, but of course they can only do so much. That’s why the publisher developed Heliograf, an in-house automated storytelling technology that specializes in short, multi-sentence updates. In other words, an artificially intelligent reporter.

One of the biggest, most exciting disruptions ever in technology, artificial intelligence is already a differentiator for winning brands. However, when it comes to publishers, The Washington Post is the exception, not the rule.

PricewaterhouseCoopers recently surveyed 1,000 business executives in the U.S. for their 2019 AI Predictions report. One big takeaway is that some verticals are embracing AI more wholeheartedly than others and media is not one of them. According to PwC, 20% of executives will deploy AI across their business this year. But only 7% of media executives are making substantial investments there.

Still, plenty of trailblazing publishers are using AI in innovative ways. Here are four of them:

Robotic Reporters

The Washington Post unveiled Heliograf in 2016 during the Olympic Games. Perfect for sports reporting, Heliograf has also authored hundreds of articles about high school sports in the D.C. area. This innovation secured The Washington Post a Global BIGGIES Award in the “Excellence in Use of Bots” category.

Algorithms can provide real-time updates about scores and results far faster and more efficiently than a person, who would realistically spend hours logging these numbers. This frees up time for people to focus more on in-depth analyses, highlighting how AI can complement, rather than replace, human journalists.

Earlier this year, Guardian Australia also rolled out Reporter Mate, to write formulaic stories on straightforward, facts-based topics such as weather updates and political donations.

…And Augmenting Real Reporters

Much like Heliograf can report football scores at lightning speed, machine learning tools analyzes data far faster and more accurately than any human. That makes AI a perfect fit for financial publications.

An AI publishing pioneer, the Associated Press tasked technology with automating quarterly earnings reports. It increased output by tenfold and with far few errors to boot. Forbes, Bloomberg and Reuters use similar tools. Unlike Heliograf, they don’t write stories; instead, they make it infinitely easier for humans to do so.

Like the AP, these publishers have AI dissect companies’ earnings instantly. They also recap numbers to provide context, suggest relevant images and make recommendations for appropriate, compelling headlines.


According to PwC’s survey, media brands are largely lagging in AI adoption because of fear and uncertainty, both about deploying AI and measuring its results. One way AI has proven to be effective is personalization. Netflix’s highly sophisticated recommendation engine cuts down on choice paralysis, increasing engagement and ultimately saves the streaming giant an annual $1 billion.

Applying that to a traditional publisher, Business Insider has more content than anyone can, or wants to, read or watch. Based on the content readers engage with, Business Insider creates profiles on them, using that data to personalize content recommendations, both onsite and in emails.

That focus on personalization helped Business Insider achieve a 60% increase in click-through rates and a 150% boost in ad click rates within email newsletters. Additionally, recirculation traffic within the site jumped 52%. The New York Times also plans to invest heavily in talent with AI, machine learning and data science backgrounds, intending to boost engagement with personalized Facebook-style feeds.

Content Distribution

AI helps Forbes journalists write content and Business Insider readers discover content. It can also help with distributing content, an area where BuzzFeed can certainly use help. Many BuzzFeed readers discover content on the publisher’s social media channels, of which there are more than 400.

BuzzFeed streamlined social publishing and took it further with AI. A joint effort between the product, social media, engineering and data science teams, BuzzFeed built a machine learning model that suggested which pieces of content to publish where, based on historic data around well-performing content.

One model helps social find trending evergreen content to repromote; another helps find similar content to repromote. A third scheduling model suggests the best publish times based on past performance.