Are machines taking over newsrooms? Not so soon
But bots are helping journalists track down news and propagate it in the most optimum manner possible
In Roald Dahl’s 1953 short story The Great Automatic Grammatizator, am an surmises that a set of mathematical principles can be used to establish the rules of grammar. He ends up creating an enormous machine that can write a prize-winning novel in a quarter of an hour. The story ends on a dystopian note with the world’s writers being coerced into licensing their creativity to the machine.
Sixty four years later, when The Economist decided to train its artificial intelligence (AI) programme on articles from its science and technology section, it wasn’t exactly attempting to invoke Dahl’s Grammatizator. It was just testing the waters for the wave of automated journalism that is sweeping newsrooms across the world.
The results were not entirely unexpected. The robot reporter managed to clone The Economist’s style and the topics they cover frequently. But even as the sentences were grammatically correct, they were incoherent. When it comes to gathering information about a subject, machines have a natural edge over humans: What techies describe as tasks that involve pattern recognition machine learning.
But they haven’t quite caught up with flesh and blood journalists when it comes to gleaning facts on the ground and getting the texture and context of the story. Where bots are indeed catching up fast – through a network of high-fangled sensors, social media feeds and advanced cameras – is in tracking down breaking news and helping journalists interpret and propagate it in the most efficient and optimum way possible. So, say a person tweets about a subway blast in a European city and the robot puts together the most relevant updates for the reporters and editors in no time, the machine is amplifying the message through technology and thereby acting like an aid to the journalist. Therefore, instead of replacing journalists, bots are actually augmenting the journalistic process. The Washington Post’s Heliograf bot which uses AI in a highly sophisticated fashion to churn out automated news reports, BBC News Labs’ news extraction tool called Juicer and Thomson Reuters’ software for machine-written articles come to mind. The New York Times uses bots to moderate user comments on its website.
One sphere where machines cannot hope to match up to human journalists for at least a few years – and thank god for that – is in thought pieces and opinion articles. They still have no answer to the power of imagination. The Great Automatic Grammatizator cannot be a match for a Dahl, the human brain that thought of it in the first place. The machines may be coming, but they are not quite here, yet.