Algorithms That Write

The New York Times published an intriguing quiz demonstrating how far computer algorithms have progressed, or failed depending on your POV, in the area of writing. You’re asked to read 8 separate passages of text and decide if each was written by a computer or a human. Sounds easy right? I did pretty good, scoring 7 out of 8 correct! Give it a try and see how you do.

It’s surprising, and perhaps a little disconcerting, to think more of what we’re reading online can and will be generated by algorithms rather than humans.
Type “robo-news” into your favorite search engine and you’ll quickly find The Associated Press among a growing number of big news organizations that have begun using robots to write stories.
But it’s not just texts being reproduced by computers. If you’ve been on Youtube lately you’ve probably come across the odd press release or news clip accompanied by a crude artificial sounding voice —robotic-like, hence the name robo-journalism or robo-news, created presumably without human intervention.

Kristian Hammond, co-founder of Narrative Science, a company producing software that translates data into “narratives”, estimates that 90 percent of news could be algorithmically generated (that is, automated) by the mid-2020s.

While this all sounds a tad dystopian let’s imagine disgraced new anchor Brian Williams as a candidate for algorithmic replacement in light of his foggy memory regarding details of the Iraq War. Curiously Mr. Williams hasn’t Tweeted anything in 5 years fueling speculation that he is not a real person but in fact a malfunctioning prototype created by NBC through a joint venture with ILM and Google’s Ray Kurzweil. Well I suppose the cat’s finally out of the bag and it’s back to the drawing board to fix Williams’ faulty memory chip.
But seriously, here’s an interesting question: would a robotic news anchor be more or less prone to making errors similar to Mr. Williams’ gross embellishment of the facts? Should Wolf Blitzer and Scott Pelley start looking for new gigs? Perhaps not just yet.

Though if this trend continues it’s not far-fetched to think journalists, and certainly many other professions, could be partially or completely displaced by advances in computer automation technology. Just as manufacturing jobs have been gradually eliminated over the years, so too writers performing repetitive work could be vulnerable to algorithms that will gladly churn out formulaic written pieces (e.g. think sports scores and financial earnings reports) faster and cheaper than most willing humans.

image credit: progue.co