Right after noon on Sunday, April 19, a reporter got furious and complained on Twitter: “I wonder how come [Helsingin Sanomat] has the latest coronavirus infection rates every day about an hour earlier than other media?”

As many others in Finland, the reporter in Turku waited for the daily data release by the National Institute for Health and then, again, she discovered the newspaper in Helsinki had already published the news. She gasped: “What about equal treatment of media?”

The answer was nothing she expected. There was no conspiracy. The article, as revealed by Esa Mäkinen, managing editor of Helsingin Sanomat, was written by a robot. 

The robot’s name is Latoja, or Typesetter. That Sunday morning, it patiently checked every minute whether any new data was available on the government Web site. The robot retrieved the figures via API, even before the government published its own charts — the source for human reporters. 

Latoja drafted an article in a few seconds, updated the charts, and then alerted its editors via Slack. “A human decides whether to publish the article immediately as it is,” explained Päivi Ala-Risku, a data journalist, in an interview with INMA. “The editor can also rewrite the draft, add some information or analysis, and publish a few minutes later.”

The robot is as good as the data it digests. In Finland, the government releases updates on the pandemic twice a day. As the morning release is shared via API, Latoja is fast and reliable.

Unfortunately, the afternoon update is published in a form of a table that sometimes changes its layout. Latoja scrapes the Web site, but silly mistakes are possible, and humans need to check after the robot.

Päivi graduated from Columbia University in New York, where she studied data journalism. A year ago, she joined the Sanomat’s data desk of 15 journalists, of which four are coders. Latoja is their baby. Five people were involved in setting up and maintaining the COVID automated coverage.

Päivi is not worried automation will ever replace human journalists: “Robots take mundane tasks, such as refreshing the government Web site’s page or copying and pasting numbers. Automation frees time of our journalists so they can focus on more meaningful reporting and analysis.”

Some routine reports scramble for automation. Latoja updates readers in seven cities of Finland with weather forecasts. Before the pandemic, it also recommended events daily, but it has not noticed the lockdown and needed to be turned off. “The truth is, if there were no robots, we would have no time to write those articles every day.”

In Päivi’s opinion, robots work best when they prepare parts of articles, such as paragraphs with updated statistics or charts. This is exactly what another robot does at Sanomat — it scans the editorial system for assigned stories and then tips journalists about charts to illustrate the topic. 

It’s up to humans whether they press the button.

Have you had any close encounters of the robot kind? Share your experiences in automation with the INMA peers. E-mail me: grzegorz.piechota@inma.org.

Banner photo courtesy of Markus Spiske and Darshni Priya MS from unsplash.

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