Washington Post, Associated Press (AP), Bloomberg, Norwegian News Agency, Forbes, Los Angeles Time, Le Monde, USA Today, Yonhap, Thomson Reuters, The New York Times, BBC, Xinhua News Agency, Press Association,… Dreamwritter, Heliograf, Syllabs, Quill, Quakebot, Soccerbot, RADAR, Kognetics, Media Brain, StatsMonkey, Wordsmith, Recount, …
The first fourteen are media, from the United States, France, Korea, China and the United Kingdom, which are already using artificial intelligence (AI) and some of its subsidiary technologies such as Natural Language Processing (NLP) to automate some processes. of the media newsrooms and free journalists to produce valuable, quality information, with analysis, research, interpretation, opinion, and thus be able to differentiate themselves from the content produced by their competitors. And, in addition, to generate new business models that include personalization, subscription and payment for access to the news.
The second twelve names mentioned are the tools used by these fourteen media, with artificial intelligence as the flag. The objective of all this is to carry out their transformation, which will also allow them to recover the essence of journalism by distributing increasingly higher quality content to readers, listeners and audiences.
Oddly enough, one of the first experiences of a communication medium to automate the production of information took place in China. Dreamwritter, developed by the video game company Tencent, wrote a 916-word business article in one minute, causing panic in the country’s newsrooms. The news, about the consumer price index, was written in Chinese, did not have a single error, and was published by the instant messaging service of QQ.com.
Today Dreamwritter, which used NLP in Chinese, has a lot of competition in the United States, especially where various tools that use natural language processing in English coexist. The Associated Press (AP) noted in January 2015 that it was automatically generating more than 3,000 stories each quarter about the financial results of major American companies, which meant increasing productivity tenfold with the same number of reporters.
In the Norwegian news agency, another example, they have been testing automation for months to write news of football results, and as their head of Innovation, Helen Vogt points out, “we are looking for new topics to test this technology”. On the same wavelength is the French daily Le Monde, which during the 2015 French legislative elections produced 150,000 web pages in four hours using an intelligent tool developed in French called Syllabs.
And it’s been two years since The Washington Post began using its own artificial intelligence technology, Heliograf, to write 300 pieces of information at the Rio Olympics. Since then, the Post has used automation to cover the news of Congress, the gubernatorial elections in Washington; and high school football games. According to the medium itself, in its first year it produced around 850 articles with Heliograf, 500 of them dedicated to the presidential elections, and which generated half a million clicks.
Less routine work for journalists
The media that use AI tools to automate and systematize some content – sports, finance, technology, motor, politics, … – also have the purpose of relieving journalists of their routine work so that they can carry out more research and analysis. and that add more value to the information. The Associated Press, for example, estimates that it has freed up 20 percent of the time from reporters covering the financial results of Dow Jones companies. And it even points out Francesco Marconi, AP Strategy director and co-head of the agency’s AI area, that “in the case of automated financial news coverage, the error rate decreased even when the production volume increased more than ten times”.
In the Washington Post, says its Director of Innovation Jeremy Gilbert, “we see potential for Heliograf to allow journalists to do their work in another way, such as detecting trends in financial data sets. We think we can help them find and write interesting stories. ” In the Post they also believe that Heliograf can help them update information, such as meteorological information. The American newspaper already has five people dedicated exclusively to making the most of its intelligent solution to automate content, “and we are still finding out what the economic impact will be, and when it makes sense to automate,” says Gilbert.
Google, which for several years has led the Digital News Innovation (DNI) initiative to help the media in their transformation process, or at least that is what it says, has provided the British news agency Press Association with almost a million dollars to create software that will collect, automate and write about 30,000 local stories a month. The solution, called RADAR (Reporters, Data and Robots), aims to “automate local reports with large public databases from government agencies.”
Thomson Reuters is also using machine learning algorithms to write stories, and The New York Times plans to automate its comment moderation. At South Korea’s main news agency Yonhap, Soccerbot is used to write stories related to English Premier League football matches. In the 2016-17 season, it produced 380 automatic experimental items, each one to two seconds after each match ended.
One final example of why I think AI and journalism are doomed to be understood, or rather are already being understood: China’s state news agency Xinhua recently announced that it is building, in collaboration with the Chinese e-commerce giant Alibabá, an artificial intelligence platform for automated news production, combining multiple data sources. It is an open platform where media agencies can apply to share their data resources. It brings together cloud computing, the Internet of Things, Big Data and AI technologies, deploying them in the news production process.
The consulting firm Gartner dares to say that this year 20 percent of the information on financial results will already be involved in automated systems with AI and NLP in the local language. And a report from the Reuters Institute for the Study of Journalism at the University of Oxford on journalism, media and technology trends and predictions for 2018, highlights that one of the most common applications of artificial intelligence is to improve content recommendations, understand the individual preferences and customize each edition in “format, time and frequency”.
And here I leave it, because there are many more examples and benefits, both for journalism, for the journalist and for the user, in the automation of information. And soon there will be more, even in Spanish. We will tell about it in Muy IA.
