“August CPI a new high in the past 12 months.”
There’s nothing remotely remarkable about the above line. But wouldn’t you sit up and take notice if told that a story with that headline was written by a robot?
Earlier this month, news sites across the world re-reported a story done by Beijing Times about robot journalist ‘Dreamwriter’ making its debut on the Tencent website.
Actually, the phenomenon of robots dabbling in story-telling isn’t new by any means. Only, more and more bots seem to be having a go at it.
Go back a few years, and you would find there has been quite a bit of press about Chicago-based Narrative Science . This is a company that promises to transform data into narratives through its software. Deloitte and Forbes are its customers. Its co-founder Kristian Hammond has predicted that 90 per cent of the news stories will be written by computers by 2030.
Just last year, Associated Press (AP) tied up with North Carolina-based Automated Insights, which has a tool that turns financial data into stories. Thousands of AP’s earning reports are being generated this way now. Yahoo! and Comcast are amongst its customers.
The low-hanging fruit of automated news is found where data becomes the basis of a story, or, as the head of communications of Automated Insights told tech site ‘Rewrite’, “anything you could put into a spreadsheet”. Stock market updates and sport, especially those of low-profile leagues bereft of any stars, are easy places to start automation.
What happens to journalists? As usual, the obvious question needs to be asked: “End of the road for journalists?” A Reddit thread did dabble with this very question in a minor way. The few answers to that ranged from the naïve (the argument that robots won’t have political bias) to the earnest (the question about how robots could do war reporting).
I don’t see journalists, especially in India, being too perturbed by the technology threat to their jobs. It isn’t difficult to look around and see some of them, even veterans for that matter, spending most of their time doing things that could make them redundant soon. That’s just because bots can do the same, or more, faster and better.
The publications that do use such software maintain they do so in order to report on areas they have neglected in the past. High school sport is an example. It has also been argued that automation has allowed the publications to redeploy journalists away from the routine and towards more critical reporting assignments.
But journalists need not be complacent. Technology is at a fairly advanced stage, and advancing further more rapidly. The tools introduced by Narrative Science and Automated Insights, for instance, can quickly spot patterns and zero in on what matters. What more, they can mimic a human writer. They are only getting better at it.
So, asking the question again, is it the end of the road for journalists? It must be said that some kinds of journalism jobs will be more vulnerable than others in the coming years. At the more secure end of the spectrum will be on-the-ground reporting. It’s hard to imagine interviewing and reporting to be anything but a human activity, especially involving those aspects of life that aren’t easily amenable to the form of structured data. However, this will be on the back of a lot of new technology which journalists have to master.
The moot point is journalists have to constantly ask themselves: what are they bringing to the table?
AP these days even has a news automation editor. Justin Myers, who has degrees in both journalism and engineering, is the first in the industry to have such a role.
More automation is inevitable. Dr. Hammond’s 90 per cent automation prediction by 2030 looks jaw-dropping. But think about it. More and more aspects of the world, hitherto neglected, will be written about in the years to come. Automation will help in the process. So, the overall volume of content will go up massively. And, therefore, the remaining 10 per cent in the hands of human journalists will still be substantial stuff.
In its blog, AP says there are far fewer errors now “than we did when we were writing earnings reports manually.” Mistakes still do occur — as in the case of a recent report that said Netflix has missed Wall Street forecasts when the company had actually performed better than expected. Some reports put even this to a human error: wrong numbers fed.
In the Tencent report, ‘Dreamwriter’ too is reported to have made a mistake in its 1,000-word piece which it churned out in a minute. It got the expert’s gender wrong. Now, such slips aren’t unknown in the world of human journalism!
sriram.srinivasan@thehindu.co.in
