Last week The New York Times ran a story by John Markoff about robots replacing human workers. Andrew McAfee, co-author of the excellent Race Against The Machine followed up with a post of his own. The gist: technology and automation lead to more job creation than job displacement in the past, but that may be changing.
Writing is one of the few areas that McAfee and his co-author Erik Brynjolfsson identified as a low risk for bot replacement. But there are a few attempts underway to train computers to do basic journalism. And while these projects can’t truly replace humans, and may never be able to, my fellow writers and I should be worried about our jobs anyway.
Narrative Science is probably the most widely known proprietor of computer generated journalism. According to Wired’s feature on the company, Narrative Science is using machines to write short news stories on Little League games and earnings statements. Apparently it’s not bad at these tasks, but it’s not like there were a lot of journalists on the Little League beat to start with.
Another example is a DARPA backed research project that tries to use bots to write dossiers similar to Wikipedia entries. The MIT Technology review reported on the project and noted that although it’s sometimes impressive, it often fails to identify key information. For example, the entry on Barack Obama listed a distant relative but left out any mention of his daughter Sasha. “Humans aren’t going to be completely replaced anytime soon,” the Technology Review piece concludes.
That’s usually how people dismiss the threat of robotized writing. But we could face the same problem other workers have in the past: more efficiency leading to layoffs. Look, for example, at how Wikipedia uses bots. Human volunteers are still needed, but far fewer than would be needed if many tasks were automated.
We write our own earnings reports here at TechCrunch, and that gives us a chance to add color and context. But what if we could have bots write the first draft, and let human go back and add detail and humanity? How many more earnings calls could one writer do if the machines took the first pass at writing the post?
Looking for stories takes up a lot of journalists’ and editors’ time. In 2010 Business Insider published the now infamous “AOL Way” memo, which described a software system for determining which stories to write about based on traffic potential, revenue potential, edit quality and turn-around time (this was back before AOL owned TechCrunch, FWIW). If I understand it correctly, editors could look at machine generated list of stories and assign them out to writers.
Tools like Parse.ly (see Sarah’s coverage) and NewsWhip Spike are designed to keep editors and journalists abreast of what’s trending on social media and could be used along the same lines as AOL’s house tools. But the thing that’s really captured my imagination is an app called Current. Here’s how Ethan Zuckerman explained it:
Zoe Fraade-Blanar presented a wonderful piece of work as her MFA thesis project for NYU’s Interactive Telecommunications Program. “Current” is a Java application designed to sit on the desktop of a journalist and monitor trending topics on Google and the appearance of those topics within Google News. The application looks for stories that have widespread reader interest (i.e., they are trending on Google Trends) and little press coverage – these, Zoe theorizes, are the stories most profitable for news organizations to cover.
As Zuckerman points out, Yahoo is already doing something similar.
I couldn’t get Current working on my PC (it’s a couple years old, and I’m not sure the APIs it relied on are still active), but it seems too general interest to be useful to me. But I wonder if Bitly’s Realtime (see my coverage) would be able to power something that could drill into niches.
Imagine a system that could find high interest, low coverage stories and generate a list of sources to contact for the story and even create a draft of an e-mail inquiry to send those sources. That would be a big boon to journalists. No, it wouldn’t fully replace a human brain looking for stories online. And it won’t be able to go out into the real world and talk to people, which is of course a great way to find stories.
Discovering news stories is actually the business that Narrative Science wants to get into, according to Wired, and CTO Kristian Hammond believes finding more stories will actually create more jobs for journalists. I’m not so sure. It will depend on a few things, like how much more efficient writers can be made through technology and how much risk publishers will take on “unproven” story ideas vs. safe computer generated ideas. The idea behind Current was that it could help publishers find lucrative stories to run to subsidize more substantial reporting. Of course publications will continue to run original, differentiating human written reporting. But the amount resources dedicated to that sort of content may change, depending on the economics of automation.
And the possibilities get weirder. Look at drone journalism. Today drones, if they are used at all, are just used to extend journalists capabilities, not to make us more efficient or replace us. But how could drones change, say, event or travel coverage in coming years? Will one reporter with a suitcase full of drones and a server full of AI algorithms do the work of three?