Hello, everyone. How has your first month of 2025 been? Mine has been busy. 

I have had the good fortune to speak to a few members of the INMA GenAI community and understand what they’re working on, and I have many more calls lined up in the weeks ahead.

Part of the reason: I’m working on programming our Generative AI for Media Master Class in April, which is going to focus on personalisation. So, if you’re doing anything interesting in that area, please get in touch: sonali.verma@inma.org.

I’m also delighted to find many inspiring examples of how AI is being used across the world, as well as to zoom out a bit and understand how to approach GenAI implementation strategically.

Sonali

15 news brands experimenting with AI in different ways

There are some really cool AI projects under way right now, at large news organisations and at small ones, from which we can all draw some inspiration.

Here’s a look at some of them:

• The New York Times built an AI tool that combed through more than 300,000 words written over 20 years to find trends, themes, and inflection points in travel. They used semantic search rather than specific keywords to better understand the content.

• The British Broadcasting Corporation has been using AI to subtitle programmes as well as to generate short-form animated sequences (rather than static images). They also provide AI transcripts of commentary for live sports and identify key moments of football matches, a popular option with fans.

“The Gen AI tools are impressive, the accuracy levels are improving, and the pilots suggest we could unlock significantly more value for audiences,” said Rhodri Talfan Davies, director of BBC Nations.

• Similarly, U.S.-based Warner Brothers is using GenAI for captioning and finding the quality of the output is actually higher than it was with manual captioning. Also interesting: They found an 80% reduction in the time taken for captioning and a 50% reduction in costs.

• The JournalismAI programme run by the London School of Economics also has some interesting projects under way:

• How could a 2,000-word article be presented — as a decision tree, interactive game, calculator, FAQs, or a summary — to help users find the information they want most? The team at India’s Scroll.in is building an AI-based tool that will repurpose news articles into different formats to share information more quickly.

• Lebanon’s Raseef22 has created Ask Aunty, a “friendly” AI chatbot for Arabic speakers to anonymously ask sexual and reproductive health questions. Envisioned as a “quirky aunt,” it offers quick answers from the news brand’s archives and trusted sources, using an informal tone.

• Ukrainian news organisation Babel faces staffing challenges amid migration and as people are called up to serve in the war. It plans to create AI avatars that replicate its journalists’ voices and appearances to allow content production to continue.

• Argentina’s Chequado is building an AI-powered assistant that helps small newsrooms debunk or validate news through fact-checking. Similarly, Spain’s Newtral is building a multi-modal AI tool that identifies false content on Telegram, which it points out is a “key platform for spreading hoaxes.” It monitors the most viral disinformation in real time, both as text and audio, and also sends alerts to fact-checkers. 

• In a similar vein, Brazil’s Gênero e Número is developing a tool to track gendered disinformation on YouTube, focusing on how influencers promote narratives that undermine women’s and gay rights, while U.S.-based Verified News Network is creating a chat product to strengthen media literacy in Indigenous communities. 

• In Nigeria, The Republic is using AI to translate content into indigenous African languages and audio, while ICIR Nigeria is transcribing audio and video into text in three major Nigerian languages. El Surti in Paraguay is working on a similar project to address the linguistic gap in AI systems for seven million Guarani speakers.

• Digitalhaus Franken in Germany is developing an AI-driven system to predict subscriber cancellation probability and recommend interventions before the customer churns.

• In South Africa, Daily Maverick is using AI to increase reader revenue conversion and helping improve reader engagement as well as subscriber retention.

What are you working on that’s interesting? Please let me know. 

Focus your AI strategy to have an impact

Like everyone else in this digital era, you now probably have many GenAI projects on the boil. How can you make sure your AI strategy makes a substantial impact on your business outcomes?

Focus on a small set of AI initiative, rather than going for breadth, and on reshaping key functions and creating new offerings, according to the Boston Consulting Group.

“Top-performing organisations follow the 10-20-70 principle. They dedicate 10% of their efforts to algorithms; 20% to data and technology; and 70% to people, processes, and cultural transformation,” the group wrote in a report.

Consulting firm Arthur D. Little has similar, complementary advice: Organisations need to remain focused on solving specific high-impact problems, not just deploying AI. 

“This means that strategic prioritisation is key, identifying where the trade-offs between data availability, AI tool capabilities and solution impact are most favourable. Make, buy, or fine-tune decisions are important — in fact, most core research, development, and innovation problems lend themselves well to fine-tuning existing open source models.” 

The critical GenAI uncertainties businesses face are those surrounding performance, trust, and affordability, the firm says. And so, businesses need to take six steps:

1. Mutualise compute power with partners to increase its affordability.

2. Encourage internal and external data sharing.

3. Better manage AI talent.

4. Train workforces in AI fundamentals.

5. Reset data and AI governance approaches.

6. Improve output controls over AI-generated content. 

Where should one focus? On hyper-personalisation, according to the IBM Institute for Business Value.

GenAI makes it possible to develop dynamic products and hyper-personalised experiences that can quickly adapt to shifting customer demands and rapidly validate changes with customers. 

“Given these game-changing capabilities, it’s not surprising that 86% of executives say GenAI is now a critical part of digital product design and development,” the institute wrote in a report titled The CEO’s Guide to Generative AI, which was released in Davos.

The way to capitalise on this ability is to redesign product development to derive high-value product insights from every customer interaction. Teams using GenAI can now conceptualise and evaluate new products in minutes, rather than days, IBM pointed out.

The hyper-personalised journeys created by GenAI could transform how companies connect with customers and employees. For example, a company can use GenAI to rapidly analyse their own customer data — as well as data from social sources and partner organisations  — to determine which customers are most likely to take various actions, from subscribing to attending an event to buying a product off the site.

GenAI can then help the company achieve true one-to-one marketing with a personalised strategy and automated, point-in-time customised offers, translated into the customer’s preferred language.

Another area of focus: “Customer service has leapfrogged other functions to become CEOs’ No. 1 generative AI priority. In early 2023, CEOs told us research, innovation, marketing, and risk compliance were the most immediate and valuable use cases for GenAI. Just months later, customer service jumped to the top of the GenAI implementation list, cited by more CEOs than any other organisational function or service. 

“This makes sense as GenAI is the next logical step for companies that have already been using traditional AI in customer service for years. In fact, 67% of these organisations have already deployed GenAI in conjunction with traditional AI in customer service. By creating dynamic, personalised experiences for both customers and human agents, this approach has the potential to spur a seismic shift in productivity and effectiveness.

“Customers want personalised answers, fast and without hassle, which means AI-powered customer service assistants are not only useful; they’re essential. But executives were particularly interested in using GenAI to assist human agents” rather than replace them — 67% were deploying GenAI for agent training, and 68% were enabling agents to interact directly with GenAI to deliver improved instant assistance.

Worthwhile links

• GenAI and the future: Bloomberg’s editor-in-chief on where we are headed.
• GenAI and deals: OpenAI is going to fund Axios’ expansion into four cities.
• GenAI and deals II: The Associated Press will provide Google Gemini with a newsfeed.
• GenAI and deals III: Agence France Presse signs up with Mistral to provide more than 2,000 news articles in six languages every day into Mistral’s chatbot
• GenAI and deals IV: Google invests US$1 billion in Anthropic, on top of earlier investment.
• GenAI and transparency: Let’s be clear when sticking a label on AI-assisted content. 
• GenAI and training data: Your video content could be worth US$4 a minute to AI companies.
• GenAI and theft: Meta knowingly trained its Llama LLM on pirated versions of books, a group of authors say.
• GenAI and agents: There are only two places where AI agents work well: coding and research.
• GenAI and education: Looks like it really works well sometimes.
• GenAI and public opinion: Looks like all Americans use AI products but nearly two-thirds don’t realise it.
• A non-AI diversion: You have probably read a great deal about the impressive gadgets at the annual CES. Please also enjoy the Worst in Show awards. 

About this newsletter

Today’s newsletter is written by Sonali Verma, based in Toronto, and lead for the INMA Generative AI Initiative. Sonali will share research, case studies, and thought leadership on the topic of generative AI and how it relates to all areas of news media.

This newsletter is a public face of the Generative AI Initiative by INMA, outlined here. E-mail Sonali at sonali.verma@inma.org or connect with her on INMA’s Slack channel with thoughts, suggestions, and questions.

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