A thesis by a software company in Chicago predicts that in 15 years, 90 percent of all online news published daily will be written by a machine. Since 2010, the company has been developing the reason behind their theory – the “Cyborg Journalist.” What at first sounds like the end of journalism as we know it, could in fact become a blessing in disguise for the shaken media industry.
The winner of the Pulitzer Prize in 2028 is already a foregone conclusion. The most prestigious award for outstanding artists of the written word will be given to 18-year-old Quill, a cyborg journalist from Chicago who, despite being a teenager, will have told more stories than Bob Woodward and Carl Bernstein put together. Is he a high-flyer in professional writing circles? Not at all. Quill is a meticulously raised narrative genius, a test-tube journalist with the narrative skills of a whole generation of journalists. But his abilities are not even based on the syntax of a normal global language.
Quill is a computer algorithm that tells stories by intelligently combining zeroes and ones. Schizophrenic? Perhaps. But already part of daily business in the editorial offices of Forbes, as well as at other U.S. publications that do not want to be named.
“How many journalists do you know who can make a living from media coverage of amateur sports?”
How Quill is able to diversify news services
Quill’s ideological father is Kristian Hammond, founder and technology executive of “ Narrative Science.” The software company, based in Chicago, developed the storytelling algorithm in 2010, giving it the code name “Quill,” and has since used it to conquer the media industry. The computer scientist from Utah vehemently rejects criticism that the service is a threat to journalism as a profession. “News written by a machine won’t cost any journalists their jobs. Rather, our service will lead to news being published that nobody reports on now.” Hammond goes on to reinforce this theory with a case from the local sports press: “ GameChanger,” a news platform that reports from local baseball and basketball games, is a service based solely on Hammond’s algorithm. Reports on local fixtures, loaded with statistics, are turned into readable prose at the touch of a button. “How many journalists do you know who can make a living from media coverage of amateur sports?” The main point for Hammond is to provide content for niches. Artificial, journalistic intelligence is no reason for cutbacks in staff at Forbes, either, but it provides them with the opportunity to distribute the latest news on finance issues more quickly and in more detail to the interested target group. This increases media range and, consequently, turnover.
The dilemma of the industry: meeting the demands of being topical with less turnover
In times where editorships are not able to meet their quality standards due to constant understaffing, intelligent storytelling algorithms offer a solution. This idea isn’t new, as Ben Welsh, data analyst at the Los Angeles Times, explains in his presentation at the Online Symposium 2012 in Texas. For years, editors like the L.A. Times have been working with “little helpers,” which Mr. Welsh describes using the archaic-sounding term “Computer Assisted Reporting.” And he explains with some pride that the software developed internally by the West Coast newspaper has twice managed to publish the fastest breaking news worldwide – which is the primary success factor in an industry that is driven by speed. After all, nothing is more obsolete than yesterday’s news. With this aspect in mind, providers like Narrative Science and its U.S. competitor Automated Insights have the answer for the rattled news industry: a service that creates fast, fact-based news in a market with shrinking turnovers. As a consequence, services like Quill shouldn’t be deemed as a substitute for real-life authors, but rather a supporting tool that helps with the time-consuming research of hard facts, their correlation, and the logical conclusions behind them. The result: a higher level of newsworthiness and efficiency.
“What distinguishes average content from stories that are told in a unique way? Humanness.”
Quality versus quantity: Does faster mean worse?
The fact that a computer algorithm evaluates statistics better and faster than a human makes sense. But what about a robot journalist’s way with words? In short: it is pretty good. Why? Services like Narrative Science were developed as learning systems. This means that each journalist from a special field feeds the robot with narrative basis material like proverbs and frequently used adjectives. Even the tone can be regulated. Whether colloquial and laidback or conservative and smooth – the style of writing is tailored to each industry and client. In theory, it sounds like heaven, but what about in practice? Lewis DVorkin, Chief Product Officer at Forbes, gives the algorithm top marks, even in normal, everyday matters. The journalist robot nails the essence of the story it’s reporting on, he says, and Dvorkin cannot think of a single failure worth mentioning. But the robot has one essential disadvantage: it is a machine. Which means that the artificially bred author lacks every emotion that a talented, real-life storyteller can work into a piece between the lines. For Matt Waite, professor of journalism at the University of Nebraska, this is the conclusive argument against content produced by a machine. “What distinguishes average content from stories that are told in a unique way? Humanness.” According to Waite, the journalism’s soul lies in the emotional undertone: hypocrisy, greed, or indignation – these are attributes that a machine can neither identify nor interpret. “This is the reason why nobody has to fear the storytelling robot,” says Waite, encouragingly. “On the contrary, we should make use of its skills. Robots provide us with more space to be human.” And furthermore, an immaculate data basis is essential for machine-produced content – something that is not a given in every journalistic field. Still, analytically driven industries, for example the finance sector, could profit on a mid-term basis.
Not only the media industry can profit
The media industry is not the only sector that benefits from writing machines like “Quill.” Narrative Science co-founder David Rosenblatt puts the mid-term development of the service in a nutshell: journalism is just the beginning, he says, with management reports in the free economy being the bigger playing field. The evidence for Rosenblatt’s thesis is his cooperation with a well-known fast food chain: sales figures evaluated monthly are summarized in short management essays and distributed to regional franchisees. This way, the contents are easier to understand and implicit conclusions are better implemented by the management. The list of potential partners for the inventors of Quill is long. From the pharmaceutical industry and the health sector to educational policy – founder Kristian Hammond sees huge potential. In the field of education, first steps have already been made in cooperation with ProPublica, a non-profit news agency in New York. Based on available education statistics, failure rates, and ethnological distribution in U.S. high schools, 50,000 single reports were produced that aim to help parents decide which educational institution is right for their children. “In 20 years, there will be no area in which a computer doesn’t write a story,” Hammond says, looking optimistically ahead. Is that something we should be prepared for in Germany, too? “The core technology of our service is language analysis, which in principle can be used to adapt the algorithm to any language,” says Hammond. “The demand is currently very low outside America, but on a long-term basis, we see other languages appearing on the horizon.”
A shortened version of this article was first published in the online edition of the German “ ZEIT ”
Text: Thomas Escher
Illustration: Jan Meyer, grasundsterne
Translation: Katharina Leonhardt & Toby Skingsley