folder icon list icon new list icon new folder Save to list notifaction icon yes tick yes tick yes tick with circle delete cross delete cross minus small - for download tool delete cross plus sign - small expander search magnifying glass icon for gettign to print page icon for email addresses icon for features timing icon for features timing LinkedIn icon Facebook icon youtube icon twitter icon google+ icon external link icon fo profile pages mail icon small mail icon for contact listings phone icon phone icon for listings twitter bird save icon export icon delete icon duplicate icon move to a diff folder mini search icon right arrow
Skip navigation
You are using an outdated browser. Please upgrade your browser.

Collaboration, experimentation, productionising and the future of AI and journalism

AI collaboration

Whether publishers and journalists like it or not, AI is having a massive impact on the media industry and will continue to do so. If you’re not already on the AI bandwagon, what should you bear in mind when using this new technology?

At a recent InPublishing webinar, fourteen publishers and suppliers shared their learnings so far with AI and what the future might hold for the industry.

Collaboration and experimentation

On the surface, implementing any AI tools may feel like a job for the tech/IT department. However, making sure the editorial team is part of the process and collaborating is essential. Aliya Itzkowitz, manager at FT Strategies, said:

‘The most important thing is involving your editorial teams or your newsroom very closely in the process. They will always understand best what your readers need and want, but then your tech teams can tell them whether that’s actually possible to build’.

Ian Mulvany, chief technology officer at the BMJ Group, backed this up and encouraged the workforce to try out the tools:  

‘The key thing is figuring out how you get people to experience the tools, allowing them to work with them, encouraging experimentation, and having the right kind of lightweight governance to make that possible within your organisation’.

It’s a point that Katja Eggert, head of strategic development at Immediate Media, agreed with and is a process they have adopted – letting teams test the product and then feedback any errors/issues or improvements.

One integrated solution vs multiple tools

The way in which companies interact with AI varies greatly, and one question to consider is whether you look for one solution or various tools to help you achieve your goals. Peter Dyllick-Brenzinger, head of product and engineering at Purple, argued that ‘having an integrated solution in your standard tools is actually quite valuable because it allows you to save some crucial clicks’. As Peter highlighted, newsrooms are already ‘very good at what they do’, and don’t need to slow processes down with extra tools.  

However, for Thomas Lake, director of product and technology at Infopro Digital, using multiple tools produced a better result for them:

‘We did a lot of testing on different specialised tools, stringing multiple tools together to really give us the accuracy we needed. When just using ChatGPT,n the accuracy was about 75%, but when we combined that more with Optical Character Recognition (OCR) tools and other large language models (LLMs), we were able to get that accuracy up to 99.9%’.

Katja also agreed that AI should be integrated into ‘connected systems’ rather than using ‘super clunky’ tools like ChatGPT. The end goal being a workflow that is completely productionised:

‘Productionising is what will really impact because it will help us achieve speech, accuracy, and consistency of scale, which is key for long-term success’.

Practical examples and the future potential of AI

Several of the publishers gave examples of how they are already taking advantage of the new technology. Tim Robinson, editorial director and publisher at National World, said:

‘We’ve found that using AI images is a way of replacing the pedestrian stock art we have used in the past, with highly arresting, highly engaging life-like images which convey the subject in a way that we weren’t able to do before’.

He did stress the importance of being open and honest about the use of AI here, and clearly marking the content. 

Tom Pijsel, VP product management at WoodWing, explained a tool that they are developing which could help editors with suggestions for shorter sentences. This could potentially increase efficiency by 50% or be faster than doing these checks completely manually. Immediate Media have been able to use AI to help with their vast archive of podcasts and to repurpose that content into other areas. It’s something that Markus Karlsson, CEO and founder of Affino, believes the industry will see more of in the future:

‘Over the next couple of years, you’re not going to have a web page with a bunch of articles on it, and people click on it and then try and find another article on that subject. It’s going to become this merger of the conversational chat and the content within it, so people can read the articles within the chat and then choose to listen to it as an audio or watch it as a video. That’s openly possible and I would fully expect that to be the dominant approach’.

Markus also believes that the emergence of Deepseek will drop the cost of other services massively. While Thomas thinks that the current ‘human in the loop’ approach will eventually turn as AI tools and LLMs become more tailored and reliable.

Want to get more information and advice on AI?  Check out our round-up of key advice from the JournalismAI Festival. Or if you’re interested in hearing from an AI expert about what changes may occur in the sector that you write about, send a request via the Journalist Enquiry Service to get the comments you need.

Subscribe to the blog
Get weekly updates from the ResponseSource blog