Automated Journalism: Reality check

Automation processes are increasingly present in our everyday lives. These processes are now transversal to professions and practices, and journalism is not any different.

Automated journalism, also known as algorithmic journalism or robot journalism, consists of news articles generated by computer programs. Through artificial intelligence (AI) software, stories are produced automatically by computers rather than human reporters. These programs interpret, organize, and present data in human-readable ways. The process involves an algorithm that scans large amounts of data, selects from an assortment of pre-programmed article structures, orders key points, and inserts details such as names, places, amounts, rankings, statistics, and other figures.

Despite it being a growing trend, there are still not that many media outlets worldwide working fully with automated journalism, so it is not deployed on a large scale, yet.

Pioneer adopters include The Associated Press, Forbes, ProPublica, and The Los Angeles Times. Early implementations were mainly used for stories based on statistics and numerical figures. Common topics include sports recaps, weather, financial reports, real estate analysis, and earnings reviews (Montal & Reich, 2016).

At the NEWSREEL2 project we tried to understand the state of use of AI, robot journalism and algorithms in the Czech Republic, Germany, Hungary, Portugal, and Romania, talking to five journalists, one in each country.


Opportunities and challenges

The world of automated journalism encompasses opportunities and challenges. For example, automated journalism can free journalists from routine reporting, providing them with more time for more complex tasks and profound reporting. It also might allow efficiency and cost cutting, alleviating the financial burden that many news organizations have been facing.

But automated journalism is also perceived as a threat to the authorship and quality of news and a threat to the livelihoods of human journalists. All this because the promise of getting more time for complex jobs such as investigative reporting and in-depth analysis of events, mentioned before, might not be accomplished.

So, in general, the main criticisms are related to authorship, credibility, quality and, looking to other activities where humans were replaced by real or virtual machines, that brings out the topic of employment.

For Rui Barros, a Portuguese data journalist, one of the downfalls for AI produced content can be that the readers feel that the text is produced by an algorithm due to “some grammatical inconsistency or some edge case that was not thought of when the product was developed”. This reinforces the need for human supervision and editing.

Steffen Kühne, from Germany, stresses out that another specific problem that tends to happen in a newsroom is linked to the integration of new software into the existing legacy systems: “Incompatible interfaces or data models are oftentimes responsible for large development and management overhead. Training and deploying AI models in a cloud infrastructure can also be quite expensive and requires in-depth knowledge on how to optimize algorithms for speed, efficiency, and stability”.

Another important note is the need for transparency, in what regards the way algorithms are used and how the automated work is done and for what is used. This is linked to  journalism and ethics, as it is to the trust in journalism.

Although there are many challenges in the present and future, the introduction of AI in journalism is growing, and projects such as JournalismAI, from the London School of Economics in partnership with Google News Initiative, intend to contribute to a wider and better use of it, allowing media organizations to explore how they could use AI technologies to approach a series of challenges, and support a growing network of almost 3,000 journalists across the world.


Reality check:  five countries, five different stages

There are substantial differences about the use and evolution of AI in journalism across different countries. To check if the perception of a global automation trends matches the newsroom reality in European countries, we interviewed journalists from five countries about AI and journalism, robot journalism and consequently algorithms: Czech Republic, Germany, Hungary, Portugal, and Romania.

Despite the differences in each country’s evolution of AI and journalism in the newsrooms, most of our interviewees agree that the use of such techniques in the newsrooms will be inevitable, so skills and literacy in these areas are crucial.

Of the five interviewees, those from the Czech Republic, Germany and Portugal report that AI was being used in their newsrooms. This is not the case in the newsrooms of our interviewees working in Hungary and Romania. But, from the five, only one of the interviewees is dedicated more specifically to work with automation, all the other have other tasks.

Most of our interviewees that work with AI use it for automated reporting, in the broader sense, for example to cover a lot of stories that happen regularly and where results can be quantified (be transformed into structured data). The most common examples quoted were reporting about sports, economy, health, weather, and traffic information. The interviewees are unanimous about the potential in using automated reporting, but some of them point out that some stories would largely benefit from human editing and context work.

In a growing area like this, having the right skills is crucial. Most of the journalists we interviewed are self-taught, sometimes reading and looking for information that helps them experiment, so it is clear that this is an area where there is a need for training. Yet, according to most of them, training programs should consist of several levels – from basic to more advanced – also because the implementation of AI into newsroom routines differs considerably between the studied countries.

In general, the importance and inevitability of AI in the newsrooms is recognized by all our interviewees, although the stage in which newsrooms are is very different from country to country. The challenges that journalism faces ahead in several areas are big, this one more, there ethical, social and more practical implications in adopting AI in the newsrooms, but the way is being paved by some media outlets.


Special thanks to Radka Matesová Marková ( Editor-in-Chief at Česká tisková kancelář – Czech Republic), Zsuzsanna Dömös (Technology journalist at 24.hu – Hungary), Steffen Kühne (Tech lead for the AI + Automation Lab at Bayerischer Rundfunk – Germany) and Rui Barros (Data journalist at Público – Portugal).

The full reports of the study are available within the NEWSREEL2 research report.

This article was written by Ana Pinto-Martinho and Miguel Crespo (ISCTE - University Institute of Lisbon).

Photo: Pixabay