Human vs. robot battles never get old. They’re usually quite fun, as is this race between a human driver and autonomous car, but they also serve as a measuring stick of the robot’s technological performance.
NPR had a fun idea and decided to put one of its reporters, White House correspondent and former business reporter Scott Horsley, to the test against WordSmith, an artificial intelligence system that uses mounds of data, quantitative analysis and some rules about style and good writing to churn out hundreds of millions of stories every year. Stories written by WordSmith are published on Yahoo! and by the Associated Press, among many others.
Both Horsley and WordSmith were tasked with writing a story about the latest earnings report for Denny’s. Horsley, by the way, is a Denny’s regular. But would that matter?
Check out the image below. It contains two stories about Denny’s, one written by Horsley and one written by WordSmith. Read both stories and see if you can tell which story is which. NPR even conducted a poll asking readers which story they liked better. We’ll share those results later in this story.
Story #1 was written by WordSmith. As you may or may not have been able to tell, it lacks the human flair of Horsley’s story. And the readers heavily favor Horsley’s story, weighing in with a 3403-270 win, as of press time, according to an NPR poll.
WordSmith, however, finished writing the story in two minutes. Scott took just over seven minutes.
Ho-hum, the humans win again. Although that could change as NPR points out.
“WordSmith’s tone is programmable. The story above was created by a program designed to mimic the straightforward tone of an AP news story. With some work from the company’s engineers, WordSmith could study thousands of NPR stories, learn NPR’s style, and start slinging its own breakfast-food metaphors.”
Listen to the full NPR report below.