Invasion-Artificial Intelligence in Newsroom

Source: BBC News Lab (

Analyzing viewer behavior, monitoring data, implementing human voice into your Google housekeeper at home… you may think this is the routine of those who work at a high-tech firm. But these are all part of BBC News Laber’s daily experiment.

As one of the world’s oldest media agency, BBC does not need to worry about their audience size. Still, it has chosen to keep in pace with the development of artificial intelligence — So how is AI re-defining the daily work of journalists, and how will it reshape the news industry in the years to come?

First, AI is changing the way journalists gather, process and manage information. What makes news stand out from other forms of media is that it provides a wide range of information with speed and accuracy. Therefore a typical task of journalist is to gather information from different sources and check its authenticity. This process usually requires people to work relentlessly on various websites to select information, put them together, and do proofreading and fact-checking — boring, slow and error prone.

And this is where artificial intelligence can play a role and make information gathering process much simpler. BBC news lab has developed a news aggregation tool named “Juicer”, which ingests text information, takes out the most important part from it(such as time, space, person’s name etc), tag them into different categories and store it in databases.

So, for example, if an investigative journalist needs to check a state-owned company and the influencing people behind it, he can use the AI tool to find all information within this company’s category including related people, and more likely, find other organizations which it has financial or political connection with. This tool will allow journalist to dig deeper into a story and see a wider picture of whatever story he’s trying to write.

Source: BBC News Lab

Juicer is the database of text. And the same logic can be applied to non-textual content such as audio, video and picture. All information, whichever media format it is in, can be categorized and made accessible to journalists in the same way. Therefore more advanced Artificial Intelligence, such as natural language processing and image recognition will have significant prospects in the news room.

Again, BBC news lab has already initiated a project in natural language processing field — “SUMMA”. This platform aims to automatically analyze media streams across many languages and extricate editorial staff from “ wasting time sifting through mountains of media by hand”. Cooperating with University of Edinburgh, BBC labbers is training SUMMA to accurately transfer audio into text even in low-quality soundtracks.

Source: BBC News Lab

Unlike text and audio, categorizing information from pictures can be a bigger challenge for AI developers. Because first, the image recognition technology needs to be smart enough to tell the difference between something similar and “exactly the same”. Second, unlike text and audio’s linear structure, the information in picture is put in a spacial and non-linear way. If AI simply tags everything in the picture, the data base would be too big and lose its aim to aggregate information; But how will AI automatically tell the important information from the unimportant? Not to mention that sometimes human information can be quite abstract, will AI be able to link a picture of May frowning with the topic “Brexit”?

Let’s put aside the tech problems and think a step further: in the near future, the databases, including text, audio, picture and even video, can then be connected together and AI can automatically put together the objects which share the same tag. For example, the text “19th CPC National Congress of China” can then be automatically matched with a news picture which shows the Great Hall of People or a soundtrack in which Xi gives a speech. And news editors can find whatever they need quickly — AI will change newsroom, as it changes everything else.

Bowen Sun

About the author: Bowen Sun, currently studying M.A. in Digital and Interactive Storytelling Lab at University of Westminster. Determined to transform digital insights into tangible solutions. Welcome to find more about me through Linkedin: