AI as Journalism Infrastructure: How News Production Is Changing
YEREVAN, March 10. /ARKA/. The most common question about the future of media is whether artificial intelligence will replace journalists. But if you look at how the world's largest newsrooms are already using AI, it becomes clear: the main shift isn't happening in the texts. It's happening in the editorial processes.
Artificial intelligence is gradually becoming the infrastructure of news organizations. Algorithms help find news, analyze data, structure information, and distribute content.
Journalism isn't disappearing. But the architecture of its production is changing.
A New Layer of Newsroom
The traditional newsroom model has long been linear. A journalist would find a topic, gather information, write a story, and hand it over to an editor. The speed of news production was directly dependent on human resources.
AI is beginning to change this logic.
A new layer—algorithms—is emerging between the information source and the journalist. AI can analyze massive amounts of data: social media posts, financial reports, government databases, documents, and video streams. It can identify potential news and structure the information even before a journalist begins working on the topic.
As a result, the newsroom can see more signals and respond more quickly.
Where AI is Already Working
In recent years, the experience of large international newsrooms has shown that AI is being used at various stages of the editorial process, from news monitoring to content personalization. Moreover, the use of AI is becoming part of journalists' daily work.
According to a Reuters Institute survey of British journalists, more than half use AI in their professional work at least once a week (another 27% use it less frequently).
Reuters has developed a News Tracer system that analyzes millions of social media posts and helps identify potential news events at an early stage.
Bloomberg uses machine learning tools to analyze corporate reports and financial documents used by journalists when working with financial information.
The Associated Press automatically generates some news stories about companies' financial results using automated data analysis systems. According to the AP, at the time of the program's launch, automation allowed the volume of such publications to increase from hundreds to thousands per quarter.
The Washington Post used the Heliograf system from 2016 to 2020 to produce short news updates during elections and major sporting events. In a more recent development, the editorial team launched the AI chatbot Ask The Post, which helps users search for information in the publication's archive. At the same time, the Washington Post is developing Arc XP, a separate technology business unit offering AI tools to other media organizations.
The Norwegian media group Schibsted actively uses recommendation algorithms and machine learning to personalize news feeds. According to INMA (2024), one test of a personalized homepage showed an overall CTR increase of more than 8% compared to the baseline version. However, the personalization logic itself is built not around maximizing clicks or time on site, but around the long-term value of the content for the reader.
These examples demonstrate that AI is already integrated into news monitoring, analysis, production, and distribution.
ARKA Agency Experience
Similar processes are beginning to occur in regional media companies. At ARKA Media Holding, the editorial team uses AI tools to monitor the information landscape, prepare new products and analytical summaries, fact-check, and process economic data. Final verification and editorial decisions are left to journalists and editors. These tools help navigate the information flow more quickly and identify topics requiring professional analysis and verification.
Key Change
AI alone rarely provides a long-term competitive advantage. Over time, such tools become available to most media companies.
The difference arises in how organizations restructure their editorial processes.
Artificial intelligence is not changing journalism per se, but the economics and architecture of its production. Media outlets that understand this early gain a significant advantage, primarily by freeing up the resources of journalists and editors to create new products and complex analytical materials that algorithms are not yet capable of producing.
In the long term, the main question for media will not be whether algorithms can write texts. Far more important is which newsrooms will be able to integrate artificial intelligence into their infrastructure in a way that enhances journalism rather than simplifies it.
The history of media shows that technological advances almost never destroy a profession, but almost always change its tools. AI will likely become just such a tool: invisible to readers, yet fundamental to news production.
Konstantin Petrosov, Director of the ARKA News Agency
