Connect with us

Hi, what are you looking for?

Tech & Science

Op-Ed: Has this article been written by a robot?

The Press Association’s headquarters in London is home to a small team of journalists, as might be expected. Perhaps more unusually there is also a team of and software engineers. The Press Association was founded in 1868 and it delivers a continuous feed of content via a national newswire, including text, images, video and data into newsrooms around the U.K. Customers include conventional media outlets, plus business brands, commercial companies, Government and not-for-profit organisations.

According to the BBC, the Press Association is developing a computer system that can perform the work of multiple human beings in selecting data trends, such as crime statistics to birth rates. The trial extends beyond the basis of data journalism to using machine intelligence to compose stories. At present these are simply written slices of news, composed of two or three paragraphs. These stories are generally the type that appear in local newspapers, aimed to be read quickly by readers who wish to obtain a quick appraisal of the ‘facts’.

This represents a new and emerging approach to news generation. Back in August 2017, Digital Journal ran an article titled “Is it possible for a robot to write the news?” This article looked into research being conducted at Jönköping University into the use of machine intelligence to create credible news stories.

At the university a project called Digital Personalization of the News is underway. The aim of the project is to see if the more straightforward (perhaps repetitive) forms of journalism, like reporting sports results or financial data, can be generated by a machine instead of a human. The success of this rests of the ability of a machine to produce intelligible sentence structures that relate to general theme in a manner that is intelligible to the reader. The research is not only orientated at the technology, it also assesses the wider social impact, such as whether the public can tell the difference between a machine generated article or one penned by a professional journalist.

In line with the Swedish university, the news agency Hallpressen (part of the Herenco group) has begun running some trial articles written by ‘robots’. Similarly, the London Press Association experiment is also generating news content. As to how good it is, the BBC has identified three examples of news content written by a machine (or mostly written by a machine, supplemented with a little human editing).

The first is from a regional British paper called The Derby Telegraph (“Shocking figures show more mums-to-be are smoking“), here is an excerpt:

The rate of smoking mothers in southern Derbyshire has increased in recent years, up from 15.1% four years ago, bucking the national trend of gradual improvement.

The news story is composed of a number of short sentences, many filled with figures. It is factual but lacks context. Moreover, the structure would not feature highly on the Flesch–Kincaid readability tests (which assess word length and sentence length). The degree to how which this matters depends on what the reader is seeking in terms of detail, relatedness and quality.

Here is a second example, from the Telegraph & Argus (“Why do Bradford households recycle less than five years ago?“):

The latest figures from the Department for Environment, Food & Rural Affairs (Defra) show that in the 12 months to the end of March this year Bradford cleared away a 231,453 tonnes of rubbish, with 87 per cent of that household waste.

And the third is from the Hereford Times (“Wye Valley NHS Trust in NHS England’s good books“):

The chief executive of NHS England Simon Stevens recently warned ministers that waiting times would continue to rise unless more money was put into the health service, after the chancellor Philip Hammond promised £350 million in the Budget to help this winter.

In other places some machine generated articles have begun to appear. For example, in 2017 The Washington Post announced it would begin publishing automated stories about high school American football matches. An editorial stated: “expanding its use of Heliograf, its in-house automated storytelling technology, enabling The Post to cover all Washington, D.C.-area high school football games every week.”

The lack of professional journalistic oversight carries risks, especially if these models are extended. For instance, automated news generation tools could be exploited by propagandists wishing to spread false news for to misinterpret facts. Simply publishing ‘facts’, which can often be contentious depending on the basis of the research methodology used to produce the data, without interpretation under serves the reader and diminishes the quality of journalism.

The extent to which the reader can tell the difference is a personal one, as is the degree to which this type of automation matters. If such stories are to become commonplace there could well be a divide between more simple news articles that contain recycled facts and quotes, generated by machines, and longer essay-style journalistic pieces that add a degree of context, cross-linkages and interpretation from the human journalist. The balance between the two could rest on how much the public eventually cares for high quality journalistic content.

Avatar photo
Written By

Dr. Tim Sandle is Digital Journal's Editor-at-Large for science news. Tim specializes in science, technology, environmental, business, and health journalism. He is additionally a practising microbiologist; and an author. He is also interested in history, politics and current affairs.

You may also like:


A digital ID is an online representation of an individual, which contains personal information.


A new phishing campaign uses HTML attachments that abuse the Windows search protocol.


Too little has been done for too long. This may well be the first instalment of the payoff.


Quick wins and dedicated transformation leaders are two ways to ensure success, HBR finds.