Every so often, a slew of hysterical articles bemoan the demise of newspapers and fret about how robots will soon take journalists’ jobs. What a crock.
I do not want to see or read about this sort of story until these fabulous computers can do optical-character recognition with 100 percent accuracy on old books. Or after voice recognition gets past 97 percent accuracy. Actually, 97 percent is not good enough. Getting three words wrong out of a 100 means a lot of copy editing, which robots can’t do either.
Most material published in a medium market newspaper is purchased in a
The best deals from the syndicate include all the news stories plus the columnists. The monthly fees vary according to market size. This method is entrenched and no robot writers are going to compete. In small markets, journalists—who may add local stories to the syndicated dreck—are already underpaid. There is no reason to replace them with a robot.
Let’s be clear that this imaginary “robot” is actually a computer and nothing more. It does not roam around the newsroom to flirt or grouse about the coffee. No, it’s a computer with news feeds going in and (ideally) articles coming out. The robot cannot do any reporting, it cannot make phone calls, it cannot go into the field. It’s just a re-write machine, so it’s a laughable idea.
It’s likely that tech journalists have thought about computer-assisted writing. Spell checkers and grammar checkers are useful, although they do not go as far as needed. None do any decent contextual editing and when they try, they suck. A finished product is perhaps 50 years in the future.
I’ve always been intrigued with computers that can supposedly tell a female from a male writer. When you run anything through a system that
It would be interesting to write an entire novel and have a computer re-write the story using Hemingway’s style. But the computer cannot do that either.
Every few years for the past 20 years, I am invited to a meeting where I am shown the ultimate robot text “analyst.” This robot can read almost anything and then take 20,000 words and abstract the whole document to two conclusive paragraphs. Demos are run. I’m convinced it works. That’s the last I hear of it. If I ask around, it turns out that it doesn’t work well enough to roll out. This means it does not work at all. That’s the end of the story, over and over.
Look at the massive intelligence agencies and all the people they employ. They generate thousands of reports. How about feeding the raw info into a computer and letting the machine crank out the reports? Why does the CIA need an estimated 21,000 employees? There are a lot of things a computer can do. Writing original material is not one of them. And it never will be.