Mobile navigation


Using AI for journalism

AI tools are already making journalists’ lives easier. Sara Forni, AI product manager at Atex, looks at how journalists are currently using AI and what lies ahead.

By Sara Forni

Using AI for journalism

Q: How are journalists using AI?

A: If we had to choose a date to mark the beginning of the boom of generative AI in everyday life and thus also in newsrooms around the world, it would be November 2022, when OpenAI made ChatGPT available to everyone free of charge. Since then, generative AI has been on everyone’s lips and various newsrooms and individual journalists have started testing the tool to see how it could help their daily work. Not all the use cases I will now list can be done using a chatbot, but it is undeniable that the advent of ChatGPT has changed the scope of this phenomenon.

At the moment, the main use cases of AI in journalism concern text production, in practices such as suggesting headlines and summarising articles, creating copy for sharing on social media and incipits and summaries for newsletters and other publications.

Alongside text production, there is also multimedia production. Generative AI is being used to create illustrations and graphics (eg. to accompany posts and carousels on social media) and more and more editorial offices are supplementing the reading of articles with audio summaries to make information more accessible. Just think how much easier it would become to stay informed if you could simply click ‘play’ and listen to the news on your way to work or while cleaning the house.

Other very common uses of AI in the newsroom are those related to audio and video transcription (again, think how much time can be saved by not manually transcribing a long interview); translation, useful both internally within the newsroom to get sources in the field (eg. international news agencies) but also externally to make the content appealing to a wider audience. Finally, several editorial offices are also testing AI for suggestions and compilation of metadata, useful for SEO optimisation of articles.

Q: What will be possible in the future?

A: In the immediate future, the first step will be to stabilise the existing use cases, adapting and customising the results for journalism. It will be necessary to move from quantity to quality and to do this, in my opinion, it will also be necessary to bring article archives and journalistic material into play. Many newspapers complain that while AI tips are useful, they are often too general and imprecise, with the risk of hallucinations always just around the corner. It will be interesting to fine-tune the LLMs by instructing them, for instance, with style guides in the field (such as the Associated Press Handbook or the BBC guidelines). With this in mind, in the medium to long term, it will be important to join forces to create an exclusive language model for journalism, as has already happened in other sectors such as finance or medicine. Crucial in this phase will be the role of journalists / content creators who, with their control and suggestions, will be able to draw up quality standards for AI in journalism. Only in this way will we be able to see an improvement in the models and thus a better use of AI in the newsroom.

Returning to more concrete applications, on the other hand, AI can be used for content planning, for instance by creating customised calendars with events to be covered and thus improve workflow. Much is already being done even now, but so much needs to be improved in the use of AI for checking of facts and statements and for the customisation of content according to the target audience (eg. rewriting a newsletter and optimising homepages according to the interests of the user who is looking at them).

Three top tips

  1. Stop believing that large financial investments are needed to adopt AI in the newsroom. There are use cases within the reach of all newsrooms and, above all, you can start with small experiments and then scale up to larger volumes.
  2. Do not implement AI just for the sake of market attractiveness and newsworthiness but implement the right solution for your newsroom. To do this, it is necessary to think in terms of the product: first understand what problems need to be solved and then, possibly, understand whether or not AI can help to solve these problems.
  3. Actively involve the editorial staff in these choices. Those who really know the day-to-day problems are the journalists who work on the news every day and not the managers who supervise the work. This will avoid the shock of having solutions implemented ‘from above’ and will also help make the content more appealing to a wider audience.

Sara and the other contributors to our AI Special took part in an ‘AI Special – Q&A’ webinar on Wednesday, 26 June. You can watch the recording by registering here.

About us

Atex offers innovative solutions for publishers: Desk for publication production, ACE for headless CMS, Insights Analytics for data-driven decisions, Kayak for subscription management, and a multi-channel booking system for ad workflows. Serving top publishers like National World and Bell Media, we make print and digital publishing efficient and impactful.


This article was included in the AI Special, published by InPublishing in June 2024. Click here to see the other articles in this special feature.