Since arriving on the scene two years ago this month, Generative AI has changed processes and possibilities for industries across the board and around the world. During the Generative AI Town Hall, presented by the INMA Generative AI Initiative, members got a closer look at the many ways this new technology is transforming the news media industry. 

INMA launched the initiative in January, and Sonali Verma, lead of the initiative, said much of the year had been spent looking at where it can create value “and where it is just smoke and mirrors.” 

Through conversations with more than 135 publishers around the world, more than 30 “Ask Me Anything” sessions with INMA members, monthly Webinars on the topic, and more, Verma has gained a comprehensive overview of how GenAI is being used.

“Gen AI has so many use cases, sometimes it feels like we’re drinking from the fire hose,” Verma said. Before bringing on a global lineup of speakers, she shared some specific use cases and best practices from 2024 and what companies can learn from them.

GenAI in the newsroom

Newsrooms have leveraged GenAI in many ways because it is “really, really good at language tasks,” Verma said. For example, Belgium’s DPG Media uses AI to edit articles by following guidelines that mimic editorial standards. This allows editors to focus on higher-value tasks, such as ensuring the content is unbiased and appropriate.

Similarly, India’s HT Media uses AI to monitor writing quality but goes one step farther, she said: “It takes a look at stories and topics that are trending and then it predicts what else the newsroom should write about that would resonate with the audience.”

These applications free up editorial staff, enabling them to concentrate on more complex judgments rather than routine editing tasks.

The power of personalisation

Personalisation with GenAI has emerged as a major trend, allowing news organisations to offer tailored content experiences. The Danish news site Ekstra Bladet allows readers to choose between full articles, summaries, or audio versions, catering to individual preferences. Germany’s Ippen also provides various formats for consuming news, empowering readers to select how they engage with content. 

“What’s significant here is that it’s up to the reader, not the editor, to make this choice,” she said. “In the past it was always the editor saying, oh, let’s do this as an infographic or as a fact box or as a proper article that will fit on this page and turn to that page.”

Argentina’s Clarin offers summaries that are 10%-20% of the story’s original length and uses bullet points, quotes, and questions and answers designed to condense and simplify the user experience.

The use of summaries is successful not just in engaging users but can improve subscription retention, Verma said. Germany’s Frankfurter Allgemeine Zeitung found that offering summaries not only improved subscriber retention but also attracted new subscribers: “They found that visits per visitor were almost twice as high for subscribers who had read summaries and that 81% of the users deemed the feature useful,” she said.

Dutch broadcaster NPO is leveraging AI by giving its audience the option to choose an “easy” version of its content. “So if the audience has a different educational background from the journalists, they can still easily understand what is being covered. It makes news more accessible, and again, the reader can personalise the way that it’s presented so that they can understand it better.” 

Amplifying audio

GenAI has made significant strides in audio content production. 

Scandinavia’s Schibsted uses AI to create continuous audio playlists, enhancing user engagement by allowing listeners to consume content hands-free. This approach has led to higher completion rates for audio articles compared to text. 

Additionally, publications like The Economist and The New York Times are monetising their audio offerings, indicating a growing market for AI-generated audio content.

Managing reader interactions

To address challenges with reader comments, OVB Media in Germany created a tool that crafts polite responses to negative comments, leading to more constructive dialogue. This tool has worked so well at calming down commenters that now it is being used to handle complaints and comments sent in via e-mail.

Packaging for new revenue streams

Japanese publisher Nikkei has tapped into a new audience segment by using GenAI to produce simplified articles on selected topics. 

“Their editors select three topics every day, and then they run AI so that the machine writes three articles drawn from Nikkei’s content on these three topics, and then they package up these three articles a day and sell it as a product for $7 a month,” Verma said. “They’ve managed to reach an entirely new monetisable segment of their audience by doing this.”

AI news presenters 

The use of AI-generated news anchors is gaining traction, and Verma shared examples from various countries, including India and Mexico. These digital anchors present news and engage with audiences on social media, and India Today’s AI anchor, Sana, has more than 5,000 followers on Instagram.

“In fact, India Today has been so encouraged by Sana’s success that it’s now created digital versions of its anchors so that when the human anchors are out in the field interviewing people or covering natural disasters, the digital anchor is on screen presenting,” Verma said. “Sometimes they have the two chat with each other to make it more interesting, interactive, and engaging for their users.” 

Ad personalisation 

GenAI is transforming advertising strategies, too, with The New York Times using it to create personalised ad targeting segments: 

“Advertisers provide their marketing briefs and then this tool uses that to build personalised ad targeting segments based on relevant articles and the audiences that engage most with those articles,” Verma said. 

New Zealand’s largest media company, Stuff, has streamlined the creation of sponsored content using GenAI, reducing costs while maintaining the quality guidelines outlined by the newsroom.

Getting better through bots

Bots can help manage some of the customer calls and provide faster service — which leads to more satisfied customers. 

In Germany, Medien Hub was getting 30,000 calls a month about newspaper delivery delays, so it built a voice bot to help its agents. 

“This bot handles 30% of calls, figures out the reason for the call, authenticates the customer based on their phone number, and provides information about the delivery situation,” Verma said. 

Humans are available for more complicated calls, but only 5% of customers want to speak with a person. If the call does need to be passed on, the bot provides the human representative with the necessary information from the conversation so the customer doesn’t have to repeat themselves.

The explosion of chat

Many news organisations are developing chat products to improve user interaction. At Times of India, an advertising chatbot gathers first-party data by encouraging users to ask questions about the product being advertised. 

“What they discovered is that there’s extremely high engagement on this; four times the click-through rate compared to traditional ad units,” Verma said. “And as a result, they can charge the advertiser a lot more for it, but also they’re collecting all this first-party data that they can then feed back to the advertiser to shape future ad campaigns.”

Chatbots have become one of the biggest trends to emerge in 2024, with news organisations around the world building their own products. Verma said these products help companies use their archives better, and they help news consumers by finding the information they’re looking for in a hurry. 

A poll of INMA members found that about 90% of respondents have either built a chat product or are building one or thinking about building one.  

What’s ahead 

In 2025, the INMA GenAI Initiative will continue looking at use cases and best practices, focusing on what has paid off for news companies. The initiative will also examine how it can be used beyond the newsroom, such as in advertising and marketing. 

“We’re also going to highlight concrete AI transformation, change management tips and tactics,” Verma said. “Many INMA members are also grappling with the question of getting the AI employees to use AI widely.

“The initial excitement really seems to have worn off. So how do we get them to not just use the tools but embrace the tools we’ve built for them?”

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .