Digital Ventures

Back to blog

Engineers Teaching Journalism



Numerous news emerges every day. As a consumer, it is difficult to differentiate real from fake news unless you have time to verify the sources.  

For journalists, it is even more challenging. They are gatekeepers who choose which news are worth publishing. As information overloads, their work becomes more difficult. So, when data management exceeds the capability of a human brain, we need…yes, we turn to technology.

On February 13, 2018, Kim Jong-Un’s brother was assassinated at the Malaysian airport, do you remember this breaking news? An aggressive assassination in public, that took some gut. This headline was on every television channel and website since the beginning.   

In Japan, before the television reported this news, there was an application that reported the event about 40 minutes earlier. You may think that this is no surprise, the television must be slower because it requires editing and verification. It is true that television news may seem “more credible” and “needs more verification” when compared to the news on the applications.

What if I told you that the owner of this application is confident that his newsroom is as credible and has a verification process that is as good as traditional media. What is your opinion?

More importantly, the developer of this application didn’t hire a troop of journalists. On the contrary, they are 24 employees, 2 in 3 of which are engineers.



NewsDigest is a popular news reader application ranking 8th in Japan’s app store. NewsDigest is a product by JX Press Corp, founded by Katsuhiro Yoneshige in 2008 when he was a freshman.

NewsDigest reports stories that fit the modern generation. It integrates the power of social media and AI. When NewsDigest’s AI discovers a “news-worthy” story on social media such as Twitter, it will search for more information before writing the news. Come to think of it, this is not different from how the new generation journalists work (which is not wrong, social media news is also news).

We may have heard about AI journalists. However, they are two separated systems. One is AI that “tracks” trending social networks news for human journalists. The other is AI that “summarizes” data that human journalists have selected (such as stocks or sports news where the input are statistics and AI provides a human language output). However, NewsDigest AI can perform both tasks, thus, it is a daily “social media news summary”.

Aside from NewsDigest for consumers, JX Press Corp has another newsroom product called FastAlert. This is a notification alert for journalists when there is a sign of a new story (journalists may call it “news sense” or “sniffing out a story”). Examples are fire, accidents, or natural disasters news that considers the credibility of the news source. At present, FastAlert users are large news agencies in Japan such as NHK, TV Asahi, and Fuji Television. TV Asahi’s executive complimented that FastAlert is a “necessary tool” because it helps journalists “sense the news” even before the police or fire stations. His interview with Bloomberg mentioned that having FastAlert is “like having one hundred million photographers”.

Is the news from AI credible? Yoneshige says that his system can filter fake news at 99%. He raised an example of an earthquake in Kumamoto when a prankster uploaded a lion image on Twitter and said that the earthquake was so strong that the lion escaped from the zoo. While social media users believed and retweeted the news, FastAlert was able to quickly verify and identify that the image was from South Africa, not Kumamoto.



At present, we often see the trend of using automated systems in journalism. Aside from products that are “alphabets”, we also see substitutions for “image” products. An example is photo stock website Getty Images (that news agencies also use) announced a system called Panel which can “replace the role of photos editors”. When journalists copy their written content onto the system, Panel will analyze keywords to find the appropriate image. For instance, in news about a power plant and pollution in London, the system will show photos of the environment in London or smog. The senior vice president of data and insights at Getty Images mentions that as of now “AI is just a tool, but editors will still need to make the selection”. The system can only select images that “seem relatable” and arrange them accordingly. It isn’t making decisions - at least for now.

Imagine a newsroom of the future with no journalist and a couple of engineers controlling the system. Can such newsroom become a reality? How can the “summary” of social media news change our world?

Surely, journalists will still be required, at least for the near future as AI still lack “judgment” and “the ability to investigate the news source”. However, new generation journalists will need to ask themselves, in this new world, what skills do they need so that they aren’t quickly “replaced”?

How can we live with AI by promoting and not replacing one another?