Menu Close
Kjpargeter/Shutterstock

Robonews – what journalists think of their new automated rivals

News writing has remained the exclusive domain of journalists until relatively recently. But with the arrival of so-called automated or “robo” journalism, this is no longer the case. Automated journalism is software that converts structured data into stories with limited or no human intervention beyond the initial programming. It has already been deployed by several large news organisations – including the Associated Press which uses the technology to write thousands of business stories every year.

Research I conducted recently with Konstantin Dörr of the University of Zurich and Jessica Kunert of the University of Munich looks at the opportunities and limitations of automated journalism. Although these are not new questions, our approach was. We recruited ten journalists from a range of news organisations and gave them hands-on training with robo-writing software from a leading supplier. We then interviewed them afterwards to analyse their views.

Speed, breadth and depth

A significant number of the journalists thought that automated news writing had the potential to reduce costs and increase speed. This was particularly the case at news organisations where there is an emphasis on fast, factual reporting – Reuters, for example. Reuters is looking at automation across the whole company and a financial correspondent there thought that algorithms were not very far from replacing 400 workers in Reuters’ Bangalore newsroom.

But the technology is also expanding coverage – by producing sports-related video packages, for example, something Reuters has not had the resources to do before.

Other journalists also saw potential in the technology to do things that were not currently being done, such as refashioning content for mobile devices and tailoring stories for different markets. For example, one journalist thought “you could automatically generate a story on a Leicester vs Liverpool [football] game and send a different version of the story to Liverpool and Leicester, and have a third one for neutrals”.

Work in progress

In spite of the projects already underway and being planned, the consensus view was that applications for the technology are currently limited. The journalists highlighted, for example, all the reporting that can’t be automated – and considered the technology a niche product.

The journalists’ views on how far the technology could replicate their skills and routines helps explain why they consider it niche. There was a significant amount of comment on the constraints imposed by the technology, in particular its reliance on single, isolated data streams and the need to predict news angles in advance. Journalists thought these constraints meant that automated journalism would lack the context, complexity and creativity of traditional reporting.

A data feed about a single English Premier League soccer game of the type that could be used to power an automated match report can contain 2,000 detailed events, from the result down to the time and position of every “on-ball event”. That data, however, relates to what happened on the pitch, which, one journalist said, might be “largely irrelevant” to the story a journalist covering a fixture might want to write. Other “big issues” off the pitch could be more important, he said, giving the example of a game between West Ham and Manchester United in May 2016.

In that match, the “riot before the game”, the fact that by losing Manchester United were unlikely to be able to qualify for the Champions League, and the fixture’s place in history as the last West Ham would play at their Upton Park ground after 112 years, were more important than the events that took place on the turf. These contextual factors, the journalist thought, “can’t be programmed in”, meaning that the data might only present “10% of the story”.

Context and consequences

Another limitation was the need to template stories ahead of time. One of the journalists asked: “Can you write up in advance something you didn’t expect to happen?” Well, to some extent you can, but, as one of the data journalists we interviewed said, to do so involves anticipating numerous scenarios, something that is “very, very complicated to achieve”.

AI will replace some, but not all, functions of journalists. Ociacia/Shutterstock

Another problem is how to provide context and interpretation. As one journalist said: “You can’t get a reaction to the data, you can’t explain or interrogate it, because you wrote the template before the data came out.”

Finally, journalists’ reactions were mixed about the ethical and societal implications of automated journalism. There were some who thought that the technology had the potential to make reporting more objective (including one from The Sun). Other journalists were concerned that the volume of media content it is possible to produce through automation could mean that the prejudices of those behind the algorithms could be more widely disseminated. In light of what we know about how social media bots were used in the US presidential election, this is a particularly relevant concern.

The human touch

While technology providers are working on solutions to some of the concerns that were identified, we believe computers are unlikely to ever understand all the nuances of human expression that help determine how events are reported. It’s also important to think about the wider societal consequences of having news and information produced and disseminated in greater and greater volumes. As one of the journalists asked: do we want as much choice as possible? In a society saturated with information, how can we work out whether something is true or not?

It may be, then, that automation will increase the need for the very human skills that good journalists embody – news judgement, curiosity, and scepticism – in order that we can all continue to be informed, accurately and succinctly, about the world around us.

Want to write?

Write an article and join a growing community of more than 180,400 academics and researchers from 4,911 institutions.

Register now