Now robots can write articles

Recently on American website Forbes articles began to appear, the author of which is specified Narrative Science (literally, "telling science"). It is a technology that allows you to create a coherent text without human intervention, and, if before the robots journalists only worked with dead statistics and wrote about sport results or financial forecasts, they can now compose and quasi political lyrics. Western authors, who now have to consider the robots as their competitors, are actively discussing the pros and cons brings to journalism this technology.

The first story, which was written by a robot, was devoted to the results of a baseball game. In most articles, which at first involved in the company, is treated either sports or financial statistics: the robot took as a basis a table or financial statement, isolate the main facts that were in the first paragraph of the article, built composition and even up to the material title, reflecting the main content.

The developers however claim that Narrative Science can work with any kind of information and any theme. To do this, the programmers can help journalists collaborating with Narrative Science. Technology Narrative Science, although it uses the same basic algorithm, modified in detail for each new customer. The easy part is the change (names, dates, fonts). The journalists put a more serious task: to determine what should be the style articles, and choose a set of specific words and expressions that you should know robot. Due to the fact that journalists ask robots the standards and language of conventional journalistic materials, articles from Narrative Science do not actually look like written by a machine.

At the moment the company Narrative Science already more than 30 clients, and profit for last year amounted to $6 million. From the construction and financial companies that used the robot to create internal reports, the technology has come in a real journalism company Narrative Science has signed a contract with Forbes and now it creates the materials for the earnings reports of companies. To call the language in the articles Narrative Science is very alive and colorful can, but the way you described all necessary facts.

The company continues to develop the technology further. In February of this year, Narrative Science has generated article on the subject: it was made on the analysis of Twitter and talked about the frequency of mentions in microblog names of candidates in presidents from Republicans. The robot made and an attempt of analysis of the data collected, although not very deep: "Although the General character tweets about Newt Gingrich, a positive public opinion relating to the candidate and his image, there is a negative trend".

But the company sets itself far more ambitious task than a simple analysis of statistics and accounts in social networks. According to the magazine, The Atlantic, the creators of technology are working on a project "audience of one", in which they hope to teach Narrative Science to generate texts intended for one person or a small group of people. The articles will be to adapt to the interests of individuals, and the subject matter will vary from estimates of the exchange offer, which stocks to buy, to the clarification of medical tests the client.

The introduction of Narrative Science in the media world, and even with such breathtaking vistas, naturally evokes a great response from the journalists themselves who see the author the robot as a competitor, and a useful assistant.

Since, as a magazine of such high class as Forbes has allowed itself to publish the opus of Narrative Science, in the press do not cease to appear scathing reviews of articles created by the robot. Some say dull and almost primitive style, others are low level analysts, and others to immediately find a consequence of low attendance articles.

Research Narrative Science, most likely (the company does not disclose details of their work), belong to the so called natural language processing (Natural Language Processing, NLP) is a theory at the junction of researches on creation of artificial intelligence and mathematical linguistics. NLP studies the problems of computer analysis of textual information, as well as the synthesis of a competent (i.e. relevant rules of a natural language) text.

I must say that the task of synthesis of the text easier task of his analysis is enough to remember how to operate machine translators or system of recognition of speech. That is why, first, the statement of the authors about the imminent creation of individual information systems should be viewed with considerable skepticism. Second, the machines work, judging by the presence of journalists, which produce patterns based on conventional neural networks, therefore, is not fundamentally different from similar systems in other industries.

Despite the abundance of criticism, and many journalists see in Narrative Science some competition at least because to create a single article, unlike the journalist, the robot takes a few seconds. In the first place from such competition may suffer the staff of the news aggregators who don't produce their own content. In addition, the computer can quickly and easily process large amounts of information the person, for example, just keep track of the constantly changing social networks. Finally, the robot, unlike a human, it is difficult to accuse of bias, and in niche media accurate presentation of facts is much more important than fine writing style.

Robot journalists, moreover, cost the owners of the media which is cheaper than ordinary people and, in all senses. Thus, according to The New York Times, one article in 500 words, created by Narrative Science, customers can pay less than 10 dollars. Moreover, the author robot frees the owner of the media on the need to provide good working conditions, pay for insurance and let the employee leave. In fact, the client only pays for the algorithm, the rest of the costs itself takes on Narrative Science.

But journalists are concerned not only practical but also ethical side of the issue. The author in an article titled "a Robot stole my Pulitzer prize", notes the possible negative impact from the technological development in the field of personalized media. He writes that because the computer will issue an article on the basis of human activity in the Internet, different people he will be teaching the same information in different ways.

So, fans of conditional Life News (or USA Today), he will issue articles, actively spiced with gossip and juicy details, and in the case that many publications will get similar robots, people simply won't be able to break out of what calls "junk press" or fast food media. In this yellow stream of articles is more difficult to survive erudite quality journalism, with which most people simply may not be able to meet you.

However, despite all the anxiety and complaints of journalists, the creators of Narrative Science argue that technology really should make life easier for members of the media. They explain that the program instantly analyzes large volumes of information (such as statistics or social media), and journalists, to save time, can work with this fish and turn it into an interesting analytical text. Thus, the technology allows you to exit media entirely new levels and achieve a completely different depth of analysis.

Of course, due to technologies such as Narrative Science competition in the media sector grows, but it grows rather in the sense that journalists have to try much more in order for their texts looked decent on the background of the materials generated by the robot. And this, in turn, may not inure to the benefit of the media sphere, which simultaneously suffocating from an overabundance of people who call themselves journalists, and the lack of really good reporters and writers.


Add new comment

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.
Enter the characters shown in the image.