The application and enlightenment of artificial intelligence in British and American news media

As early as 2014, the Associated Press began to cooperate with Automated Insights, a company that provides automated writing services, to let the robot Word Smith read financial reports and then automatically generate financial news with a bias towards data analysis according to the structure provided in advance by the editor. This artificial intelligence system also has the function of "self-learning". It mastered the basic norms of news writing within three months, and the lower error rate also improved the quality of articles. The next direction of the Associated Press's artificial intelligence is to automatically convert text reports into voice broadcasts.

The Washington Post has developed nearly 100 "intelligent robots" so far, the most well-known of which is Heliograf, which debuted during the Rio Olympics. It can automatically generate stories based on real-time data sources, use software to search massive amounts of information to help reporters dig out exclusive news points, and provide readers with personalized stories. In the Rio Olympics report, Heliograf obtained the latest information about the event from sports data companies and automatically edited it into short messages as instant news releases. On the day of the 2016 US election, it tracked the voting results and automatically generated emails and tweets. In 2012, four reporters from the Washington Post spent 25 hours manually editing and publishing some of the election results, while in 2016, Heliograf created 500 reports alone, with almost no human intervention. The Washington Post plans to use Heliograf to continuously update information produced by humans or machines. For example, if someone shared a report from Tuesday on Thursday, Heliograf can automatically add the latest facts to the report. Heliograf will even have the potential to write news, with journalists only needing to provide relevant facts, analysis and opinions, and the robot will then complete the synthesis of the articles.

USA Today uses Wibbitz artificial intelligence production software to generate short videos. It can compress news reports into a video script, string together selected images or videos, and even add commentary.

The cooperation between Reuters and Graphiq has also been quite fruitful. By using the intelligent algorithms of Graphiq's huge visual database, you can pre-hypothesize events to be reported and allow artificial intelligence to quickly capture relevant data and match them with the reports, which means providing relevant data for your press releases and saving editing time.

Reuters is also developing a sports reporting tool that can, for example, automatically generate slideshow-style short videos from NBA game images, with commentary extracted from the "automatically generated three-paragraph report" by this tool.

The social network classification monitoring system provided by NewsWhip of the Associated Press assists in reporting decisions. News Whip has been engaged in news reporting monitoring since January 2014. Currently, the social networking sites it monitors include Facebook, Twitter, Instagram, Reddit, Linked In and Pinterest. The data monitoring frequency is once every two minutes, and the data content covers 30 languages ​​in 100 countries. It is said that through algorithmic analysis, the company can predict 79% of major hot news, telling users which content is most widely spread among global audiences and which reports are the most effective, so that the editorial department can make more effective decisions on reporting topics. Reporters and editors of the Associated Press have received News Whip training to monitor breaking news in their own regions and the languages ​​they use.

Thomson Reuters' algorithmic prediction tool can help journalists judge the authenticity of a tweet. It can monitor possible news based on the "credibility" and "news value" of the tweet. Its judgment basis includes who is tweeting, how the tweet is spread on the Internet, and whether there are users on Twitter trying to confirm or falsify the sudden event.

Buzzfeed, an American news aggregation website, has also launched a chatbot, BuzzBot, which was originally designed to report on the Republican and Democratic National Conventions in the United States. It can collect information on the scene and has now developed into an open source reporting tool that relies on robots. Unlike other news agency robots, Buzz Bot does not provide services such as news recommendations and online food ordering. Instead, it collects news materials from delegates, protesters, and anyone on the scene through interactive communication with users to help Buzzfeed news website produce reporting content. Users can send photos or other information directly to Buzz Bot, and Buzz Bot will occasionally ask users questions or ask users to submit details related to their own experiences.

The New York Times' Blossomblot can perform big data analysis based on articles on social platforms, infer which types of content are more popular, and assist editors in selecting appropriate materials. According to the newspaper's internal statistics, the number of clicks on articles screened by Blossomblot is 38 times that of ordinary articles! In addition, Blossomblot's machine learning capabilities can also provide illustrations, summaries, and even set titles.

CNN is one of the first users of Facebook Messenger chatbots. CNN launched a chatbot that focuses on personalized distribution and pushes headlines to users every day. Below the daily push, there are three options: Read Story, Get Summary, and Ask CNN. Clicking on the three options in turn, users can read the story content or get a summary of the story, and can also ask questions to the CNN chatbot.

The New York Times developed NYT Politics Bot, an intelligent robot based on the personality and charm of one of its political reporters. The robot uses his voice to report daily presidential election situations to readers, enhancing the closeness of communication with readers and creating a one-on-one interactive experience with readers.

Judging from the practices of American and British media, the application of data science and artificial intelligence is bringing great convenience to the production and organization of news media, but it is also posing a number of adaptive challenges.

1. How to enhance one's own technical reserves and innovation capabilities, build and maintain a database with sufficient and accurate content, and ensure that the news generated by the algorithm is correct. Building and maintaining a database with sufficient and accurate content is the basis for algorithm-generated news. Only by ensuring that the original data manually input into the algorithm is accurate can the news generated by the algorithm be correct. The algorithm itself also needs to be continuously improved, which requires higher technical reserves and innovation capabilities of news media. The US and British media generally adopt a combination of self-development and outsourcing cooperation.

2. How to transform the business process of news gathering, editing and publishing to seamlessly connect with the artificial intelligence system. The introduction of artificial intelligence systems will first change the production methods of primary news products such as stock market and foreign exchange market financial reports. The news gathering, editing and publishing process and job arrangements will inevitably need to be adjusted, and even the business sector settings will need to be changed. The functions and manpower requirements of the editorial department will naturally need to change. The intelligent reporting system itself needs to be maintained and monitored, especially experienced editors are required to input reporting templates into the machine, update the templates, monitor machine errors, and ensure that the workflow becomes better and more efficient.

3. How to strengthen the technical training of traditional editors and build a new generation of editors and reporters who are more adaptable to the AI ​​reporting system. The new generation of reporters is not only "omni-media" in the general sense, but also "cross-border", not only good at reporting, writing, editing and publishing, but also good at data analysis and computer technology, understand how the AI ​​system works, and how to use the AI ​​system reasonably in reporting. (Source: China Media Technology)