Could a robot write this column?
That's the question that popped into my head when a colleague forwarded me a story on National Public Radio about a company called Narrative Science that developed software that can take data and spit out news stories. In one case, NPR reported, the robot wrote a terrific story about a college baseball game, while a live person who covered the game for a sports site called Deadspin buried the fact that the pitcher pitched a perfect game.
When I first became a reporter, I found it frustrating that my employers seemed happiest when I produced stories I considered robotic. Any idiot could go to a city council meeting, or a fire, or a car accident and report what happened. I didn't need a master's degree for that. The publisher never seemed to appreciate the stories that took more thought and energy and creativity, although those were the stories that seemed to resonate most with my readers.
The scary thing for most journalists is that given the standard facts that go into any news story, a robot could churn out the who, what, where, and when of a story. But it's the how and why that gets tricky.
Much of our local media seems to reflect robotic reporting. Reporters take facts spat out by some public information officer at city hall or the police department or university. But it turns out that a robot can do that just as well, or better. A robot knows the rules of grammar. So where does that leave the journalists of tomorrow? Out of jobs?
Here's the thing. Every new development in journalism makes me more excited about the field. I say let the robots do the robotic reporting. I never wanted to do those stories anyhow. But I doubt a robot could find news that hasn't been released. It couldn't fight public officials when they claim that documents are part of the public record. You wouldn't catch it saying to a public relations manager or mayor: "C'mon now, you expect me to believe that garbage?'" Can a robot go to three different people who each have three different accounts of the same event and figure out what actually happened? Facts are one thing. Truth is quite different. What happens when the facts as released aren't true?
Forget that. Can a robot go to a chili cook-off and describe the difference in taste between Joe Smith's rabbit-and-kidney bean chili and Helen Taylor's with venison and pinto beans?
A number of scientists in Silicon Valley are hard at work developing artificial intelligence systems now that can do more than process facts. They are working on computers that understand humor. They now write programs that teach computers how to recognize puns and master the timing of jokes to develop jokes with punchlines. Apparently it is all math. But just as there is a difference between being able to process facts and figure out the truth of a situation, there is a difference between being able to come up with a joke and being just plain funny. Some of the funniest things in the world are unexplainable, like a rubber chicken or an arrow through Steve Martin's head or a pie in a face.
On the opposite end of the spectrum, I don't think a computer could handle things that are just plain tragic. Can a robot know when to express outrage? Would it know when to not report facts when the reporting could harm a family and the information isn't something readers need to know? Could a robot have the sensitivity needed to interview family members who lost loved ones in a tragic accident or disaster or crime?
Part of the craft of journalism is source development. No robot could wheedle facts out of a police detective reluctant to disclose them or build up enough trust with a whistleblower to be able to expose corruption or mismanagement of public funds.
Now that I think about it, wouldn't it be great if there was some type of software that would expose corrupt politicians or bureaucrats or police officers? To intern at the district attorney's office in Humboldt County you need to first take a lie detector test. Imagine if all politicians had to take one before they got on the ballot. It would never happen. That's because "truthiness," as Stephen Colbert would say, is part of how we all operate. I teach my daughter that honesty is so important, but still she knows that if people ask her how they look, she should say, great! even if they look god-awful. A robot can spit out facts. But can it tell a fact from a sort-of fact? The truth from the sort of true? Can it read between the lines of what people say and catch the unspoken truth? And can a robot help me understand a world that seems to make no sense?
In a classic Star Trek episode Captain Kirk, Spock and Bones are trapped by a bunch of androids (all in the form of beautiful women). Kirk figures out how to disable the androids with what's known as the liar's paradox, the statement "I always lie." The robot goes into overload because if Kirk is lying, he is telling the truth.
I say bring on the robots. Any way to get more information out to the public faster and easier is a good thing. That will only free up the good writers and proactive reporters to find the stories that would otherwise go unreported and to question the "facts" that the robots spit out. I'll still be there to say to the robot: "C'mon, you expect me to believe that garbage?"
Marcy Burstiner is an associate professor of journalism and mass communication at Humboldt State University.