For the past several years robots have quietly infiltrated newspapers and magazines, slowly replacing reporters. Sounds like the plot of a “B” movie? Maybe, but robots have learned how to write like humans, how to do research on the internet, and are becoming a part of the regular publication process. There are robots among us, and they are taking over… our jobs! Now, the takeover is about to go into high gear!
The idea of robot writers has been around for decades. After all, many forms of popular literature follow simple formulas. Take romance novels. A college professor, Robert James Waller, taught his students that romance novels follow a simple formula. Waller wanted to prove this to his students and spent one summer writing, “The Bridges of Madison County”. It became a top selling romance novel, and later a top grossing movie.
A year later (1993), the next obvious step was taken and a computer wrote a romance novel in the style of Jacqueline Susann, a top-selling romance novelist. The book was readable but never became a best seller. Nor has any robot written a top-selling book. However, computers have been successful at writing news stories. For example, PR Newswire writes thousands of short (2-3 paragraph) stories about how well publicly traded corporations performed this quarter. They look something like, “Company X has released its quarterly financial results, declaring a dividend of $0.15, exceeding analyst predictions by 20%”, and so on.
Before this task was automated, PR Newswire had three problems. First, reporters hated this work; it was boring. Second, while thousands of publicly traded companies need stories written every quarter, reporters are often called away to work on more urgent stories; not all financial reports were written (or written on time). Third, when writing repetitive stories that have a lot of numbers, human writers tend to make a lot of mistakes and need to write retractions. Robots took over 2 years ago. Now 100% of the stories are written, on time. With no mistakes. Just in case you’re wondering, no editor has yet managed to hurt a robot’s feelings with last minute assignments, or boring tasks.
Strangely, robots have been amazingly successful as sports writers. They have also solved an old problem. There are always more sports events than reporters. National newspapers can cover major games, but no newspaper has enough reporters for EVERY game. Not just major and minor leagues. What about college or even high school teams? How much coverage do women’s professional sports get? Emerging sports, such as eSports (electronic gaming) are ignored. Yet, eSports have huge cash competitions (up to $20 million grand prizes), sell out stadiums (45,000 seats in Seoul, South Korea) and have massive on-line viewership (32 million viewers for a recent competition).
Robot reporters are so fast and cost effective, they can write stories for “micro” audiences too small to justify a human writer. Consider a high school volleyball team. All of the team members, faculty, friends and family might total just 100 readers. No professional publication could justify the cost of a reporter (and publication costs) for this audience. Yet, a robot could cover every game.
If the mother of a volleyball player reads the New York Times, she would tag her subscription with her interest in local stories about volleyball. Larger events (a college game?) might show up in the custom print version of the NYT, and the high school game might would show up in her digital version. Robots can provide this customization all news, not just sports. Local political news, economic issues, weather, and other stories that impact just a small group that are ignored today could be written by robots. Robots can also run user forums and write custom letters to answer individual user questions. Robots can make the news more personal.
Newspapers and magazines are desperate for some way to stop the slide in readership and revenue. Customized news may be their best hope. It so happens that two enormous opportunities have arrived for robot writers… the 2016 Olympics and the Presidential Elections!
I’m writing this blog just after the close of the 2016 Olympics. If you closely followed the games, you may have heard about the Washington Post (WP) using robot writers, from it’s homegrown “Heliograph” project. Heliograph robots will write simple stories, mostly Tweets, and social media, supplementing their team of sports writers. No one expects any of these stories to win a Pulitzer prize, but by relieving their reporters of the duty to issue Tweets, Heliograph allows human reporters to make better use of their time, on higher value stories.
In the last few years, social media has often been humiliating for reporters. Professional reporters are supposed to be first on the scene for news stories, but some random guy with a cell phone usually gets out the first Tweet, Facebook comment, or Pinterest photo. Of course, that “random guy” might misspell a name or report the score wrong. A professional reporter has to get it right the first time… even when a delay of a minute or two means that you send out Tweet #1,000 rather than Tweet #1.
Speaking of sports events that don’t get enough coverage, the Paralympics will start in Rio on September 7th. The Paralympics is the world’s biggest international competition for athletes with physical or mental disabilities. It receives far less coverage than the Olympics, even though there is as much drama and as many record-breaking performances.
Consider the infamous Oscar Pistorius, a legless runner whose top speed in the 100m sprint is just a second behind Usain Bolt. Looking for human interest? The Olympics banned nearly the entire Russian track and field team. In the Paralympics, EVERY Russian team has been banned. Other athletes have faked disabilities in order to win gold medals. The Paralympics has incredible stories to tell. There just aren’t enough reporters to tell these stories.
Of course, if you live for scandals, few events can stand up to the Presidential Election! Political coverage faces the same challenges as sports. Every publication will cover the Presidential race. Senators? Sure. Congressmen… yeah. But state politicians? City officials? When the elections end, how many reporters will be left to cover local political beats? At most newspapers today, the answer is… unfortunately… just about none.
Associated Press and Forbes use robots to produce quarterly financial statements. The Washington Post is all-in on robotic reporting. Forbes Magazine printed at least a few software generated articles. A reporter at the LA Times created, “Quakebot”, to monitor for seismographic data and automatically generate stories after each earthquake. The L.A. times printed at least one Quakebot story. Interestingly, Quakebot was based on an earlier L.A. Times “bot” that writes stories on homicides.
The L.A. Times’ parent company, TRONC, owns The Chicago Tribune, Orlando Sentinel, Baltimore Sun, Sun Sentinel and Hartford Courant. released a video that explains their new digital business model. They plan expands the use of robotic reporters and uses Artificial Intelligence (A.I.) to research stories, and manage marketing. Machine Learning will be used to analyze the behavior and reading habits of consumers, increasing engagement. A.I.’s that provide real-time analysis of reader data could easily move from supporting editorial decisions, to making them. Could a robot editor be in the works?
Imagine that a TRONC A.I. manages editorial and marketing decisions. Editorial decisions would be data driven, market data driven. The A.I. monitors trending social media stories, reads competing publications, predicts which stories will be popular, polls readers, and determines where there are (profitable) publishing gaps.
Editorial decisions and marketing would be different sides of the same coin. TRONC’s video is clear that their FIRST stories need to pay the bills; only after expenses are covered will their newspapers have resources to write other stories (that address social justice, or win a Pulitzer, etc.). If robots don’t already work for your favorite publication, they will be soon. Print and news media need something new to lower their costs and stop the erosion in readership. Robots will not just cut costs, they might bring back readers. The number of reporters will shrink as robots take over the management of social media, but research, marketing, and even editorial functions are under just as much cost pressure and will also see some automation.
Wouldn’t it be ironic if the headlines about the robot revolution… will be written by robots? Probably! One thing is sure, more of the content you see in newspapers and magazines will be researched, written and marketed by A.I.s. There will be displacement in employment, but we may also enter a golden age of customized news. That’s my Niccolls worth for today, and I’m sticking by it!