Why is AI the best ally for journalists?
by Lawrence Carbon, CTO of ContentSide
How does this technology impact the daily life of journalists? Let us remember that the press is one of the sectors strongly impacted by the development of digital technology. On the kitchen floor, the entire information production chain has been transformed, from collection to distribution, including page layout and proofreading of articles. On the reader's side, the Internet has become the preferred means of accessing information.
This development has not finished shaking up the economic models of press companies. Although they have of course integrated this new situation into their production chains and, with varying degrees of success, into their business models, the use of the latest generation of tools based on artificial intelligence is not yet widespread. The reasons are diverse, fear or lack of knowledge about this technology, integration perceived as complex and expensive in the business ... This, while AI is becoming one of the major levers to help journalists focus on the essential.
Facilitating the production of new content ...
In the daily life of journalists, AI-based natural language analysis facilitates the identification of articles on the same subject. It also accelerates contextualisation by facilitating access to ad hoc sources, where searches based on keywords or traditional indexing are not necessarily relevant. For example, a journalist working in the regional daily press on Georges Brassens will more often search for the name of a street or a cultural centre, while his colleague working for the specialist press will search for the latest papers published on the author-composer.
Searches are made all the more complex by the fact that the volume of articles produced on the Internet has increased significantly. Indexing such a volume of documents is too big a task for documentalists, and would also require a complete harmonisation of classification methods. Because it is capable of recognising and extracting complex entities without constraints on the volume of documents to be analysed, AI answers these questions.
It can also facilitate, or even automate, the production of new content, particularly when the sources to be taken into account are numerical. The articles generated by AI simply "textualise" these figures and allow journalists to concentrate on analysis. This technology is also used for distribution. It optimises natural referencing by identifying key concepts and increases reader engagement via personalised bounce links.
... without heavy investment
All of these possibilities stem from the latest generation of algorithms used for natural language analysis. These technologies have made a quantum leap in the last 4 years and interpret each word in the context of the sentence and text in which it appears making the difference between "...four times the annual gross margin..." and "...violent behaviour by a few bullies on the margin...". Also, the necessary computing power is now very affordable and can be used on demand via the cloud. Based on these developments, SaaS offerings are developing at affordable rates.
At the same time, specialised "ready-to-use" solutions drastically simplify the implementation of AI functionalities and avoid launching in-house projects, a risky approach that requires specific skills that are difficult to find. In this way, the weekly magazine Le Point integrated a suggestion engine into its editorial flow in less than three months, using a pre-trained AI solution.
Increasingly simple and economical to integrate into research or content production tools, these AI-based functionalities are strengthening editorial strategies and transforming the business models of press titles. AI facilitates the daily life of editors by correcting, contextualising and extending searches, thus freeing up their time to deepen their analyses. The press of tomorrow will certainly be composed of "pens" combined with ... AI.
* Lawrence Carbon is CTO of ContentSide, a French ESN and software company specialising in AI applied to content management.
 Transformers: https: //lbourdois.github.io/blog/nlp/Transformer/
Are you interested in this topic?