Don’t fear the robot reporter

Don’t fear the robot reporter

Voor de master New Media & Digital Culture schreef ik onderstaand essay over robotjournalistiek.

Last summer, the journalists in the newsroom of the American business magazine Forbes welcomed a new colleague: Bertie. Their new co-worker was able to provide them with real-time trending topics to cover, recommended ways to make headlines more compelling and suggested relevant imagery for their articles. However, Bertie is not a human being, but an innovative content management system  named after the magazine’s founder B.C. Forbes. According to Salah Zalatimo, Chief Digital Officer at Forbes, Bertie’s revolutionary artificial intelligence gives Forbes’ storytellers a “bionic suit” and he promised that the magazine will “regularly roll out new AI features to further augment our storytellers’ natural abilities” (Zalatimo 2018).

            However, Forbes’ Bertie is not that revolutionary at all. In 2014, the Los Angeles Times was the first newspaper worldwide to publish a story about an earthquake thanks to the Quakebot, an algorithm that automatically generated a short article when an earthquake occurred (Oremus 2014). Since the same year, news agency Associated Press has been using the services of Automated Insights, the creators of Wordsmith. This robot platform is able to convert data from reports into news stories. The agency has published thousands more stories thanks to the technology (Automated Insights 2019).  

            Over the last decade, journalism has had reasons to adapt and innovate for several reasons. To name the elephant in the room: we currently are living in a time where full-time editors and reporters are the victims of layoffs at digital publishers and traditional media chains (NVJ 2019). Most news organizations have actually reduced staff, requiring the remaining employees to multiply their output for the 24-hour news cycle and multiple platforms (Van der Haak et al. 2012, 2924). Journalists have been finding ways to cope with this situation and journalism generated by machine is on the rise.   In order to cope with the reducing amount of newsroom staff news corporations have been steadily adapting algorithms, programming them to introduce automated news writing. 

Andreas Graefe explains in his Guide to Automated Journalism that algorithms can “create thousands of news stories for a particular topic, they also do it more quickly, cheaply, and potentially with fewer errors than any human journalist” (Graefe 2016). But, what exactly is this form of robot journalism or automated journalism? According to Matt Carlson, author of The Robotic Reporter, automated or robot journalism is the algorithmic process that converts data into narrative news texts with limited to no human intervention beyond the initial programming (Carlson 2015). This lack of human intervention leaves questions about the agency in journalistic work, algorithmic bias and future of journalistic labour. As a freelance journalist myself, I feel the possible effects the rise of robot reporters may have on my own career and when I think of this I feel scared. However, I have put this opinion under scrutiny and although these concerns are substantial, in this essay I would like to argue that we should not fear the robot journalist. And I will explain you why.

New technologies » caution

But let us first take a broader view and look into the ways new technologies have been approached and how they already are affecting and may affect the way we view jobs. Since the beginning of civilisation, mankind has been sceptically watching new technologies develop. For instance, take the words of the Greek philosopher Plato when he argues that the creation and adoption of writing will lead to nothing good. In his work Phaedrus Plato (370 BC) lets his character Socrates say:

Since the nature of speech is in fact to direct the soul, whoever intends to be a rhetorician must know how many kinds of soul there are. […] Words that have been written down can do more than remind those who already know what the writing is about? […] Once been written down, every discourse roams about everywhere, reaching indiscriminately those with understanding no less than those who have no business with it, and it doesn’t know to whom it should speak and to whom it should not (548 & 552).

 According to Plato, the then new technology of writing changed the relation between humans. When we speak directly to each other, he argued, we directly transfer our soul and feelings to our conversation partner. We have direct influence on our own words and this changes when we write our words on paper, Plato argues. The text we write will spread autonomously outside of the influence of the writer. In this evolution, the new technology of writing creates its own agency separated from human influence. When considering Plato’s Phaedrus as an early media theory, one can argue that new technologies should be approached with care.

            It is easy to approach robot reporters cautiously with Plato’s words in mind, because they will change the way we practice journalism just as much as the transition from speech to writing has influenced the way men have communicated. Currently, we have no idea what the labour market will look like in 2050, but it is clear that algorithmic processes and automated learning will alter nearly all our current jobs (Harari 2018, 38). However, this definitely is not only a bad influence.

Humans ¹ jobs

For every job that was altered or taken over by machines since the Industrial Revolution, at least one new job for humans was created (Mokyr et al. 2015, 36). However, this statistic is based on two main human capacities: our capability to perform physical and cognitive jobs. For example, when physical labour was being automated in the agricultural sector, we could take up jobs in the service sector. We could deploy our cognitive abilities to analyse, communicate and understand human emotions in these jobs (Harari 2018, 39). This much praised “human intuition”, however, is simply a matter of pattern recognition, Daniel Kahneman states (Kahneman 2011, 26). 

Pattern recognition is a trait that is learnable for any self-learning algorithm. We humans may simply be deemed not essential for certain jobs in the future. Certain jobs that require routine activities may easier be automated, as they require a clear input and output (Harari 2018, 46). A short news article, for example, has a fixed structure of writing and the news can be collected from a selected number of news sources. Graefe describes the most common form of robot journalistic technology as “simple code that extracts numbers from a database, which are then used to fill in the blanks in pre-written template stories” (Graefe 2016).

A good example of how this trend is already affecting journalism is the aforementioned Quakebot of the Los Angeles Times, the first successful robot reporter. The Quakebot created short news stories about earthquakes in the region by itself, replacing journalists whom had been producing the same kind of news articles (Oremus 2014). Pessimistic journalists may stop here and wait until their jobs are there to be taken over by other on algorithms based robot reporters. But there is more. The abovementioned view on robotics and journalism is very dualistic and may be based on the binary reasoning that algorithms will take over our jobs and all journalists shall be unemployed by 2050 (or maybe even earlier). This is a wrong assumption from my point of view. In the next paragraphs, I will explain why and also provide a set of recommendations for future approach and collaboration with robot reporters.

Robot reporters + human journalists = one goal

Up until this point, this essay has described the dualistic aspects of the emergence of robot reporters. It has addressed possibilities for human journalists having a job or losing their job to self-learning article-writing algorithms. It has also focussed on the fact that robot reporters are different from their human counterparts. At this moment in time, they lack human emotions and are only programmed to “fill in the blanks in pre-written template stories” (Graefe 2016). However, I believe that we should move away from these dichotomies in Latourian fashion and create a new vocabulary and understanding of our relationship with the robot reporter algorithms.

            The French philosopher and sociologist Bruno Latour has extensively tried to transgress the distinction of human and non-human actors (Latour 1991, 111 & Latour 2005, 75). He is one of the leading theorists behind Actor-Network Theory (ANT), an approach that is used to challenge our biases of the unequal distribution of power across relations between human and non-human actors, forming networks of hierarchy, rejecting dualistic notions (Latour 2005, 191). The theory describes a social world that is formed by connections between actors, who can be human and non-human. Its goal is to redefine what the “social” is. The actors are defined by an ability to act, to perform and participate in the process of making a difference (Law 1992, 379). They are aided in this process by actants. In The Politics of Nature, Latour writes that actants are anything that “[…]modif[ies] other actors through a series of […]” actions (Latour 2004, 75). Latour proposes a horizontal ontology in which there will be no distinction between actors and actants.. ANT, he writes, “claims that there is nothing specific to social order; that there is no social dimension of any sort, no ‘social context,’ no distinct domain of reality to which the label ‘social’ or ‘society’ could be attributed…that actors are never embedded in a social context and so are always much more than ‘mere informants’” (Latour 2005, 4). The “social” is made up of connections between actors and actants and it is only when there are these connections, one can speak of any semblance of the “social”. One of Latour’s classic examples of Actor-Network Theory is that of a gunman. Latour illustrates the associations made between human and non-human actors which create the “social”, when he – according to Riske (2011) – states:

Here it is stated that a man and a gun can form a new entity when they are connected in a third entity: the gunman. […] A man cannot shoot someone all by himself. However, it cannot be said either that the gun is the cause of all problems. Guns that shoot someone all by themselves are quite rare. The connection that ANT wants researchers to focus on is the connection that brings the man and the gun together, and thus creates a gunman. A gunman is different from both a man and a gun in the sense that a gunman is able to shoot someone whereas both the man and the gun cannot do this alone.

In this example, both the human actor (the man) as the non-human actant (the gun) is needed for an action of shooting to take place. According to Latour, both actors and actants are needed to create the gunman and shape his actions which take place in the horizontal network created between the actors. However, Latour never discussed the phenomenon of artificial intelligence and robot reporting or automated journalism in the context of ANT (Mlynar et al. 2018, 4). Therefore, ANT does not fully account for the degree of agency in these new journalistic developments. But, I believe that the human journalist and robot reporter should be seen as separate parts of a new entity: the journalist of the future and I propose that we see robot reporters as actors that along with journalists share the goal of bringing news that is as factual, quick and well-written as possible.

Labouring/Creative journalist

Robot reporters can be the answer to current problems for journalism as the spread of fake news and mistakes made in online news media due to budget cuts for example. Automated journalism creates new opportunities for newsrooms as it could augment or algorithmically ease the burden on human journalists. There is the opportunity for freeing human journalists from repetitive tasks. This will also cause a shift in the vita activa, as coined by Hannah Arendt (Arendt 2014, 394). This German philosopher and political theorist proposed labour, work and action as three “fundamental human activities” for creating the active engagement of people with the world (Arendt 2014, 389 & Parekh 1981, 103). The act of labour has to be fulfilled in order to provide for the basic necessities of life (Arendt 2014, 389). Work involves the construction of artefacts, it is a purposeful activity that stimulates our creating capacities and skills (Arendt 2014, 390). In this activity, we are not only the toiling animal laborans (Arendt 2014, 401), but a creating homo faber (Arendt 2014, 397). Action entails communicative activity and in particular, political speech in the public forum (Arendt 2014, 390). At this moment in time, the work of a journalist can be places in all of Arendts categories. This however changes with the robot reporter. With the robot reporter taking over repetitive tasks as writing short news articles with a fixed format, a part of the labour activities of the human journalist are taken care of. The human journalist is then freed of writing simple repetitive pieces. This will leave human journalists with more time to concentrate on background stories and story-telling behind articles.

Currently, the most common fields of use for robot reporters among European news agencies are finance and sports (Fanta 2017, 10). Both fields feature clear input for the algorithm (financial data and sports results). One of the first adapters to this kind of automated financial content was Bloomberg News. Their program, Cyborg, wrote thousands of articles last year that analysed financial reports and turned them into news stories like a business reporter (Martin 2019).

But robot reporters can also detect discrepancies in a given data set, a skill used by journalists of LA Times to identify mismatches in the actual number of crimes in their city and the crime stats of the Los Angeles Police Department (Nahser 2018). This information generated by robot reporters resulted in creative content and thus benefiting journalism. In the first wide-ranging survey of automation in European news agencies, Fanta argues that in the future the robot reporters will be used more to identify stories. “Algorithms will find interesting patterns or outlier events in the data and flag them up for journalists to look at. […] This might take the form of automatically generated statements on the data, to be followed up on by journalists looking for explanations and reactions” (Fanta 2017, 17).  Also, more jobs may be formed as robot reporters need maintenance and enhancements to the algorithm. With these possibilities and future potential for robot reporters, I argue that journalism should claim that the relationship between human journalists and robot reporters shall not be defined as one of objects and subjects. We are in Latourian reasoning fundamentally relational to each other – the robot reporter is an assemblage humans have created to lessen the burden on human journalists and ultimately with them teaming up, creating a better journalist than before.   

There are, however, elements concerning robot reporting that have not been elaborated on deeply. I would like to recommend these topics for further discussion on an academic and journalistic level. Firstly, automated technology in journalism is still in its infancy. Journalistic fields where structured data is available like sports and economics are currently the main focus for robot reporting innovation, as news agencies are reliant on data providers. News agencies of the future should distinguish themselves from the competition, and also take notice of data integrity where missing items of data can lead to bias in content generation (Dörr & Hollbuchner 2017, 413). Also, journalist-driven innovation likely will still require improving journalists’ digital skills, empowering them to take charge of more aspects of digital innovation (Fanta 2017, 16). Therefore, more digital literacy will be needed for future journalists. And thirdly, as objectivity, responsibility and accuracy are journalistic values of media organisations, the need for algorithmic transparency is key in gaining trust for robot reporters (Diakopoulos and Koliska 2016).

To conclude, in this essay I have argued that we (being: humans) have to move away from dualisms when it comes to the relationship between robot reporters and human journalists. We have to view them as colleagues re-enforcing each other on an equal level, both working to benefit journalism. We should not fear robot reporters and automated journalism as they provide us more journalistic tools for content-creation and fact-checking, more focus on our creative qualities and the creation of jobs for people monitoring and tinkering the robot reporter algorithm. So, do not fear the robot reporter, as its rise will benefit journalism.


Arendt, Hannah. “The “Vita Activa” and the Modern Age.” In Philosophy of Technology: The   Technological Condition: an Anthology, 389–405. Chichester, West Sussex, UK; Malden, MA: Wiley Blackwell, 2014.

Automated Insights. “Associated Press.” Accessed May 17, 2019.

Carlson, Matt. “The Robotic Reporter.” Digital Journalism, 3:3 (2015): 416-431,

Diakopoulos, Nicholas & Koliska, Michael. “Algorithmic Transparency in the News Media.” Digital     Journalism (2016). DOI: 10.1080/21670811.2016.1208053

Dörr, Konstantin Nicholas & Hollnbuchner, Katharina. “Ethical Challenges of Algorithmic Journalism.” Digital Journalism, 5:4 (2017): 404-419.

Fanta, Alexander. “Putting Europe’s Robots on the Map: Automated journalism in news agencies.”        Reuters Institute Fellowship Paper (2017).

Graefe, Andreas. “Guide to Automated Journalism.” Tow Center for Digital Journalism. Accessed        June 19, 2019.

Harari, Yuval Noah. 21 Lessons for the 21st Century. London: Jonathan Cape, 2018.

Kahneman, Daniel. Ons Feilbare Denken. Amsterdam: Business Contact, 2011.

Latour, Bruno. Reassembling the Social: An Introduction to Actor-Network Theory. Oxford, Oxford University Press, 2005.

Latour, Bruno. Politics of Nature: How to Bring the Sciences into Democracy. Cambridge,        Massachusetts, Harvard University Press, 2004.

Latour, Bruno. “Technology is Society Made Durable”. In John Law (ed.): A Sociology of Monsters:  Essays on Power, Technology and Domination. Sociological Review Monograph No. 38 (1991): 103-131.

Law, John. “Notes on the theory of the actor-network: Ordering, strategy, and heterogeneity.” Systems   practice 5.4 (1992): 379 – 393.

Martin, Nicole. “Did A Robot Write This? How AI Is Impacting Journalism.” Forbes. Accessed June     21, 2019.

Mlynar, Jakub & Alavi, Hamed & Verma, Himanshu & Cantoni, Lorenzo. “Towards a Sociological       Conception of Artificial Intelligence.” (2018). 10.1007/978-3-319-97676-1_13.

Mokyr, Joel, Vickers, Chris and Ziebarth, Nicolas L. “The History of Technological Anxiety and the      Future of Economic Growth: Is This Time Different?” Journal of Economic Perspectives 29:3 (2015): 33 – 42.

Nahser, Freia. “Three examples of machine learning in the newsroom.” Global Editors Network.           Accessed June 21, 2019.

NVJ. “De journalistieke arbeidsmarkt in vogelvlucht.” Accessed May 17, 2019.             

Oremus, Will. “The First News Report on the L.A. Earthquake Was Written by a Robot.”      Last modified March 17, 2014.

Parekh, Bhikhu. “The Vita Activa and the Vita Contemplativa.” In Hannah Arendt and the Search for    a New Political Philosophy. London: Palgrave Macmillan, 1981. 

Plato. 370 BC. Phaedrus. Translated by. A. Nehamas & P. Woodruff: 548 – 556.

Ritske Dankert. “Using Actor-Network Theory (ANT) doing research.” Accessed June 20, 2019, 

Van der Haak, Bregtje, Parks, Michael & Castells, Manuel. “The Future of Journalism: Networked        Journalism.” International Journal of Communication, 6 (2012): 2923 – 2938. 

Zalatimo, Salah. 2018. “Entering The Next Century With A New Forbes Experience.” Forbes.   Accessed May 17, 2019.

Geef een reactie

Het e-mailadres wordt niet gepubliceerd. Vereiste velden zijn gemarkeerd met *