Artificial intelligence has been a hot topic in the news for a while now, with plenty of related requests being sent through the Journalist Enquiry Service by the media. But how is AI affecting the publishing industry itself?

The JournalismAI Festival, celebrating five years since it began, delved into topics including personalisation, fact checking and storytelling techniques. Here are the key takeaways from the panels:

Journalist and ‘expert’ guidance for AI

Several speakers throughout the sessions stressed the need for AI tools and processes to have a ‘human’ in the loop. 

Alessandro Alviani, generative AI lead at Süddeutsche Zeitung Digitale Medien, said his team are looking at a different approach – putting an ‘expert’ in the loop. This change came about following a mix up made by a prototype summarization tool, which claimed that footballer Pascal Gross transferred from Brighton in the Premier League to the German national football team (not true):

‘This summary was reviewed and approved by an editor. Why? Because this editor is not a football expert. It sounds grammatically and stylistically like a perfect summary, but it’s wrong. That’s the reason we need an expert.’

AI for the media industry performs best when it is guided by journalists, and ideally journalists that are experts in their field. Peter Andringa, visual investigative journalist at the Financial Times, shared the paper’s direction: ‘the underlying thread for all of our reporting is that it’s human-led and AI-enabled. We see AI as one tool in our toolbox for reporting these stories.’

Policies, training, and AI literacy increasing

More newsrooms are incorporating AI into their workflows, and just as journalists must abide by editorial standards and guidelines, there is a need for policies around AI. Gemma Mendoza, head of digital strategy at Rappler, said ‘you need to have a policy, and that’s one of the things we started with’. 

It was the same for Darla Cameron, interim chief product officer at The Texas Tribune:

‘Our first step in the process with AI was to set some guardrails for our work. We developed this policy and the first line was that we will not use AI to replace our journalists, but we do encourage cautious experimentation.’

Companies are also focused on training journalists to use the new technology and making sure that they are AI literate. This is something that Nikita Roy, data scientist and journalist and host of the Newsroom Robots podcast, has spotted:

‘What we’ve seen in the industry this year especially has been a very proactive approach towards AI literacy and integration, which I think has been promising. It’s not just been isolated experiments here and there. A lot of the larger newsrooms, and even smaller newsrooms, are systematically moving towards understanding and leveraging AI.’

Benefits of personalisation

According to Trustingnews.org, 94% of people want journalists to disclose their use of AI. Readers may be wary of AI, but they are more likely to trust the news when they are told how media companies are using it. There are also benefits for them when it comes to personalisation. Mukul Devichand, editor of audio programming at the New York Times, explained how Snippets (part of the NYT app) uses an ‘algotorial feed’ – a mix of personalisation algorithms and curation by editors. 

Khalil Cassimally, head of special projects at The Conversation, has started experimenting with content creation using AI and a more personalised approach for different types of users, including creating a micro site for younger audiences around the elections in Indonesia.

Many outlets are taking steps towards greater personalisation for readers and turning to a more conversational dialogue. But Clare Spencer, product manager at Podmorph, believes that the potential of generative AI can not only help with personalisation, but format too:

‘I’m quite excited about the possibility of the audience being able to consume the content in the way that suits them, rather than in the way that we find it easiest to tell them. An example from this year is that WhatsApp rolled out the feature where they transcribe your voice notes. That means the person sending the message sends it in the most convenient way for them, and the person receiving it can receive it in the way more convenient to them. I think for news that is really exciting.’

The future of journalism and AI

While AI is already having an impact on the newsroom today, the final panel of the day discussed what the future holds for journalism and this new technology. Nikita Roy believes that we will see an evolution into an AI-first newsroom: 

‘We’re heading into this AI-first era where anybody can produce content, and they are able to receive content and understand information in the way they want, in the language they want, and in the specific locality in which they want. An AI-first world will force newsrooms to start thinking about not AI as just a theme, but as an infrastructure you’re building within your newsroom.’

Charlene Rolls, editorial director of Media 24, thinks that subscription and advertising will be transformed completely as media companies find different sources of revenues with AI. Clement Baccar, head of data and AI at Brut, summed up four main changes he expects to see with ‘vocal assistance, new models for video generation, more assistance in general, and more specialised open source models and LLMs for companies.’