The phrase “robot journalism” starts off problematic with many newsrooms, yet many media companies say it is worth the risk.

During an INMA members-only Webinar on Wednesday, United Robot’s Chief Marketing Officer Cecilia Campbell said that while robot journalism is “an increasingly popular way to expand the capacity of the newsroom,” she acknowledges the term itself is somewhat problematic.

First, the word “robot” can be confusing for reporters. “It’s not robot vs. human,” Campbell said. “It’s not an either/or choice.” With very few specific exceptions, United Robot’s media partners have not used their technology to replace reporters — in fact, a few have even hired more journalists after deploying the robots — but rather, they’re “using automated content as part of an investment in better local journalism.”

Second, the word “journalism” can be misleading. “To put it simply,” Campbell said, “journalists do the journalism. The content produced by our robots, it’s not really journalism — it’s information.”

Campbell discussed five ways that robots can add value to a newsroom.

1. Volume: “The volume of text they produce means you get wider, more granular, more hyper-local coverage.”
2. Speed: Robots can write a story in seconds, immediately as something happens. What’s more, it takes them as long to write one story as it takes to write 100.
3. Consistency: Having regular and reliable data, which is what United Robots builds text on, means you’ll be able to publish current and updated information.
4. Accuracy: “Robots don’t make mistakes. If it’s in the data, it’s in the story.”
5. Story discovery: Robots analyse data. It’s how they identify stories. They’ll pick up on unusual events or patterns, and these insights can be used as alerts for the newsroom so reporters can either follow up on a story or create completely new stories. “That’s where robots can help produce journalism.”

Newsroom concerns, automation risks

One of the big concerns publishers express about robot journalism is how the newsroom will react. Campbell emphasised it’s important to stress that it’s not an either/or proposition, but rather that they complement one another really well.

Some of the risks United Robots has seen with content automation are:

• The biggest risk they have seen is the impact of inconsistent, irregular, or incomplete data, Campbell said: “The content is only as good as the data.”
• Using a self-service tool to build your own robot creates a dependency on in-house tech experts.
• There’s potential editorial imbalance depending on how and where automated content is published. It’s important to not “flood parts of your site” with automated content, but instead to distribute it thoughtfully.
• If it’s done badly, it could damage your brand.
• It can be difficult to get the newsroom on board, but it’s important to get buy-in from everyone in the organisation.

Two publishers who are clients of United Robots then shared their experiences with automated content.

NTM addresses 3 concerns of robot journalism

Jens Pettersson, chief digital reader revenue officer at NTM in Sweden, said they’ve been using robots since 2019. About two years ago, they started an initiative to increase subscription numbers and have been using robots to help achieve that goal. The robot is fast, has the capacity to produce a large amount of content, and it’s focused solely on what we want it to cover.

NTM has been using the robots to cover the topics of real estate sales, sports results from lower divisions, traffic news, and newly started companies. They’re looking to expand that list to include weather and business reports in the near future.

Pettersson discussed how they maximise the effects of the robots by mixing human reporting with robot reporting.

“We have an algorithm that runs our front pages,” Pettersson said, “based on news value [as determined by] the editors, plus things like subscriber page views and number of comments.” Because the real estate sales content is extremely popular (the more a home sells for, the more interest there is from subscribers), they can “set a threshold value so that the higher home prices will appear automatically on the page.”

In addition, though every single home sale is published individually, they can use the information to create aggregated pieces on topics like the most expensive homes sold during a particular period of time. And these pieces are “smash hits for engagement,” Pettersson said, often “among the most read articles of the day.”

The robots are only reporting on facts, on numbers. But then journalists can take a deeper look at the data, Pettersson said, digging into, for instance, why houses in a certain area are selling for higher prices.

In terms of numbers, Pettersson said NTM’s most-read reporter had an impressive 4.9 million pageviews from subscribers last year — while the robot had 9.4 million pageviews in the same year.

Pettersson also shared the three big concerns that NTM found:

• Organisational: Will they steal our jobs? The answer here is a clear “no,” Pettersson said. The robots will help increase productivity on specific topics, on the pieces “anyone could write,” leaving reporters time to “focus on the stuff that demands thoughts and consideration.”
• Editorial: Are the texts and headlines of high enough quality? Pettersson said this answer is “yes — mostly.” The quality improves with time. Today, the text they see is more varied and has fewer mistakes than when they first introduced the robot, though he echoed Campbell, saying that “the only time something is simply wrong is when the input data is bad.”
• Business: Is it worth the money? The answer to this question is, Pettersson acknowledged, at this point is: “We’re not sure. We hope so, but the calculation is quite tricky.” He said they know the automated content is a great addition to all their local journalism, it’s much cheaper to produce than if a human had to write each piece, and the pieces contribute well in terms of conversions and engagement. On the other hand, however, it requires a significant investment of time to optimise the use of the content to get the most out of it.

NDC Mediagroep shares 4 reasons the risk is worth it

Ard Boer, product owner of NDC Mediagroep (now owned by Mediahuis) in The Netherlands, discussed how they’ve been using United Robots to dramatically increase their coverage of football matches around the country.

In a country of about 17 million people, more than one million participate in the amateur football leagues, Boer said. There are hundreds of matches played every week. And while they send reporters to cover the big matches and elite leagues, the vast majority of smaller local games got no coverage at all.

With the robots, Boer said, “We can report those match results — all of them — to help stay connected with the community.” It wasn’t hard to implement, either, since they’d already been receiving all the data from the Dutch Football Association in real time. It was simply a matter of turning that data into robot-generated content. They make sure they don’t just publish the match results but also things like how those results impact league standings and win streaks. This makes the information more useful and valuable to local subscribers.

Boer said using robots isn’t without challenges, both external and internal, but he shared the four reasons why it’s worth the risk for them:

1. “We can report on events we never would have covered in the past, where it’s not worthwhile to send a reporter.”
2. “We can produce more hyper-local content,” and though it’s of high interest to only a limited audience, more hyper-local content means those limited audiences can add up quickly.
3. “We can produce unique content,” since no other media outlets are covering these events — for the same reasons they used to have.
4. “We can publish articles in almost real time.”

INMA members can watch the replay of the Webinar here and see the schedule of future Webinars here.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .