Relevance

The number of media organizations that automated journalism providers currently report as customers is small. Few providers offer actual journalistic products, and most products available to date are limited to routine topics, such as sports and finance, for which reliable and structured data are available. Automated journalism is thus still in an experimental or, at best, early-market expansion phase.³⁶

This may change quickly, however. Apart from ongoing advances in computing power, big data analytics, and natural language generation technology, the most important driver of automated journalism is the ever-increasing availability of structured and machine-readable data provided by organizations, sensors, or the general public. First, in an attempt to make government more transparent and accountable, many countries are launching open data initiatives to make data publicly available. Second, our world is increasingly equipped with sensors that automatically generate and collect data. Currently, sensors constantly track changes in an environment’s temperature, seismological activity, or air pollution. Sensors are also increasingly used to provide fine-grained data on real world events. The NFL now uses sensors to track each player’s field position, speed, distance traveled, acceleration, and even the direction he is facing—which provides many new opportunities for data-driven reporting. Third, users are generating an increasing amount of data on social networks or among parents at local youth sporting events.

Furthermore, automated journalism fits into the broader trend within news organizations to commercialize journalism and follow business logics. In light of declining profits and readers’ increasing demand for content, news organizations are constantly looking for new revenue and production models that help cut costs by automating routine tasks and, at the same time, increase the quantity of news. Due to its ability to produce low-cost content in large quantities in virtually no time, automated journalism appears to some researchers as yet another strategy for news organizations to lower production costs and increase profit margins.³⁷

Given these drivers, it is not surprising that advocates of automated journalism expect the field to expand quickly. Saim Alkan, CEO of the German software provider AX Semantics, estimates that already today algorithms would be capable of producing about half of the content of a regular daily newspaper. Alexander Siebert, founder of Retresco, another German company, thinks that within five years automated news will be indistinguishable from human-written news.³⁸ And Kristian Hammond, co-founder of Narrative Science, predicts that within the next ten years more than ninety percent of news will be automated.³⁹

These claims are certainly debatable, in particular as they come from people with a vested interest in the success of automated journalism. However, with renowned news organizations such as the Associated Press spearheading the movement toward automated news production, it is likely that others will follow suit. Lou Ferrara predicts that “every media outlet will be under pressure to automate” and, eventually, “everything that can be automated will be automated.” Similarly, Tom Kent, AP’s standards editor, expects an “explosion of automated journalism.”

In fact, there are indications that more and more media companies are already heading in this direction. Most providers of automated journalism solutions are in constant negotiations with media organizations interested in their products. Narrative Science and AX Semantics declined to provide information about journalistic clients, as non-disclosure agreements prevent them from revealing existing collaborations.⁴⁰ Still, automated journalism might already be more common than is publicly known.