Can robots write news? Turns out they already are. In the Wired article “What News-Writing Bots Mean for the Future of Journalism,” Joe Keohane explains how Heliograf and other artificial intelligence are writing data-heavy stories about sports and election results.
Heliograf is owned by Washington Post, and Washington Post is in turn owned by Jeff Bezos, the CEO of Amazon. Bezos bought the Post in 2013 when newswriting AI was in its infancy. In 2016, Heliograf got its start auto-publishing articles on the Rio Olympics.
After some tweaking, editors wrote key phrases for potential election results, like “Republicans retained control of the House” and “Democrats regained control of the House,” plugged Heliograf into election data from VoteSmart.org and set the AI loose on the election.
Heliograf can choose phrases from the template, fill in data and publish multiple articles across different platforms. If there are anomalies in the data, such as wider margins than expected, Heliograf can alert reporters.
Keohane is cautiously optimistic about AI news writers. AI is suitable for local news, like Iowa election results, when local reporters don’t have the time or money to cover stories. The winners in this case are the Post and other news sites that have AI technology, because they are growing their audience reach while saving money not paying reporters.
Heliograf is also updating both robot and human-written stories as new data comes in. In this way, AI can free reporters from mundane stories to investigate other news and save news companies money so they can afford to keep reporters rather than downsizing.
If AI supplements, rather than replacing reporters, reporters win and so do readers. Readers would benefit from better news stories at lower cost from both AI and reporters who are free to cover more nuanced stories.
In the best-case scenario, the losers are smaller local news companies that cannot afford AI technology. Local news is already struggling financially without factoring in other companies using AI.
In the future, I don’t see AI ever fully replacing humans. AI can help lower the cost of news and this relates to the concept that media is primarily funded by advertising. So, with advertising revenue decreasing, AI could help save news companies and create new jobs when it replaces mundane ones. For example, we don’t have telegraph operators any more, but now a whole new field of IT has opened up.
For now, Heliograf is reporting on objective, factual stories, so there is not a lot of agenda setting going on. It’s possible that robots will improve enough to write opinion pieces.
In the future, robots and big data could craft political stories that slant left or right for specific audiences.
This article did not cover some obstacles that could happen. AI could make a mistake and report false news. Readers who read fake news and think it is real, would lose, as well as the news site affiliated with the gimmicky AI. (The L.A. Times’ Quakebot, which is AI that reports earthquakes, accidentally reported an earthquake that happened in 1925 as if it happened this year, according to an article by Novack).
The article also did not talk about transparency. I can’t tell if articles by Heliograf say they were written by a robot, or if they bear the name of a staff member. The Post should disclose to people when the articles they are reading were written by a robot, but it doesn’t look like they are right now.