Skip to main content

ACCUEIL > Nos actualités > > How is AI helping journalism?
How is AI helping journalism?

How is AI helping journalism?

Programs are made to solve problems and automate tasks: a source of risk for some, and an ingenious solution for others, several sectors will be affected by Artificial Intelligence (AI).

However, some have more difficulty in getting involved and adapting. This is the case with journalism.
Knowhere News uses the Artificial Intelligence slightly differently. Firstly, it is used to identify the most popular topics of the moment on the web. The pre-selected sources are all verified sites. Nevertheless, they are not prioritized in the same way, according to a confidence index: the AI will thus give more weight to an article from the New York Times than one from Infowars.

To avoid the propagation of fake news, Knowhere has chosen not to take into account the statements made on social networks and the AI only chooses a topic if at least five verified sources have already published on this story.

Next comes the automatic writing phase of the "No Human Bias" article, which focuses on facts corroborated by several publications. This process now takes between one and fifteen minutes. Two editors then proofread the post and improve the grammar (which in turn allows the algorithm to progress). 

What new journalistic tasks are made possible by AI?

Detection of media information: Firstly, it is used to identify the most popular topics of the moment on the web. Once the topic is selected, the AI will scan over a thousand different sources and create a database for the topic in question.

Fake News Detection: Fake news stories are not new, but the problem is not so much that people no longer trust the media, but rather that they trust any fake news. Tools using AI can generate them in a few seconds by computer: Verification of information (authenticity check, metadata analysis, comparison of information with the database) 

Again, the program is not the silver bullet – most initiatives and tools work in combination with humans, whose ability to analyze and verify sources, even just by making a phone call, still exceeds that of robots. To optimize search, programs can easily be trained with content click rate data. This technique does not work for the detection of fake news. Here, the datasets to train the programs to detect fake news must be coded by human fact-checkers.

But what about the Journalist in all this?!

Many journalists nevertheless fear this technology. And naturally, they don't want their work to become obsolete. Human journalists will continue to be needed. In the long term, writing algorithms may allow journalists to refocus on writing reports and interviews – forms of writing that involve more subjectivity.

Today, journalistic bots are capable of reproducing information sorting, search and selection techniques. However, they are not yet sufficiently mature to match the writing style of a human journalist! Nor are their analytical, reflective, specific objective setting and information cross-referencing skills.


Régie Numeryx