For those still operating in newsroom silos, heads up: AI calls for interdisciplinary collaboration, advises Uli Köppen. On strategy, too: “It’s the most important thing here” – because you’ll have to relook your infrastructure and data management systems, for sure. But finding the ‘talent gold’ in your newsrooms, and building interdepartmental talent pipelines, and teams will boost your organisation’s entry into AI. And once you get started, the possibilities are, well, endless.
She should know.
Köppen is an award-winning journalist at the forefront of AI exploration in newsrooms, with a solid track record in developing crossover tech/journo teams, having mapped out the route herself.
She joined German public broadcaster Bayerischer Rundfunk (BR) as a freelance online newsroom reporter in 2019, was soon promoted to Team Lead Cultural Content – still as a freelancer – and, in June 2012, joined BR full-time as Head of Cross-Media Journalism.
Since then, she’s been instrumental in building up several teams for BR and, today, leads three: Data, Investigative and BR’s AI + Automation Lab, “an unusual combination between content and product” that comprises an interdisciplinary team of journalists, coders, and product developers.
Here, she works on prototypes in product, and contributes to investigative algorithmic methods, “feeding learnings back to digital strategy”.
“What these teams all have in common is that we’re using algorithms for journalism – investigative journalism – and for product,” explains Köppen.
A 2019 Nieman Fellow and participant in Online News Association’s Women’s Leadership Accelerator 2022, Köppen is also renowned as an international speaker on journalism, automation and strategy, and shared her invaluable experience with audiences at WAN-IFRA’S Newsroom Summit 2023.
Pioneering tools to strengthen digital strategy
Cultural barriers pose the biggest challenge to implementing AI in newsrooms, says Köppen, emphasising the importance of understanding workflows and using existing tools and channels for easier transition.
For example, using a language model to filter comments and integrate them into the team’s communication channel, making it easier for newsroom staff to find relevant comments and engage in discussions: “We built a simple alert system for the newsroom to alert them to the right comments, they just get a link that they could click on, which leads them to the CMS,” she explains.
BRk is clearly ahead of the curve when it comes to tech exploration and advances, from the organisation’s structure, to its output. According to Reuters’ latest annual Changing Newsrooms Report, only “29% of news leaders report that their organisations already have in place high-level principles to guide their AI plans.”
BR’s AI and automation Lab’s mission is clear: “We’re using AI and automation to serve our users better. Of course, we’re doing quality journalism and AI and automation should help us to improve that, and to answer fragmented user needs better. But as public service broadcasters, we don’t want to just copy Netflix and Amazon; we want to find public service ways of personalisation,” explained Köppen.
Last month, the public broadcaster rolled out its personalised regional audio prototype.
Okay, this is a big one for us Say hello to our latest AI-supported news product release:
“BR Regional-Update” creates your own personalized regional newscast on demand. https://t.co/SfZXbs92VJ
(Currently available for Bavaria)
— BR AI + Automation Lab (@BR_AILab) October 25, 2023
“We consider these projects as Trojan unicorns, because they are bringing technological change for us. We’re not only interested in the product, even though we think the product is nice, and a good idea; what we are really interested in in this case is the new metadata infrastructure we are building behind it: versioning, and regionalisation.”
Versioning, explains Köppen, works on the idea of getting the news according to your needs: “according to the time of the day, according to device, and according to your confirmation habits.”
An example of this is Summaries, which are being rapidly adopted by various news organisations, and the Covid automation BR released during the pandemic.
Now, they are looking at prototyping video text right with newsroom workflows, “because we want to build workflows around tools like that, and then we want to adapt the tools to build workflows.”
“With this metadata infrastructure, we can personalise audio as we want, by using other tags, for example, persons or topics. Of course, we had the technology in our lab before, but the fact that everyone can put their hands on large language models and try it out themselves, makes us faster.
“We’ve been working for years on our prototype, but now we’ve become faster, and reduced our prototyping cycles to a few weeks.”
Köppen attributes BR’s tech advances to their dedicated AI Expert Board, which started about five years ago with interdepartmental meetings, that led to an AI strategy, and guidelines that have been in place since 2020. ”This board is crucial, because everyone’s on the same page,” notes KUoppen.
“We know what the others are doing; it’s not hierarchical at all. It’s a great way to find silos and then try to break them up.
“You want to dive more into interdisciplinary work.”
This, really, is the crux of Koppen’s presentation: a lobbying call for interdisciplinary teams in newsrooms to investigate AI and automation.
“I think this is a very important task right now, for us as journalists and also something the audience appreciates, because there are a lot of questions on how AI works; on how it changes our society, and how we, as a society, want to use AI and automation. Where are the good sides? Where are the dark sides?”
The investigative work produced by the AI and Automation Lab may address some of those questions. Beyond a selection of stories exploring algorithms in society – such as the German credit scoring algorithm Schufa, a vaccination algorithm, and how platforms are using algorithms in different ways – they’re also using AI as a tool.
“In an ironic twist, we used an image recognition algorithm developed by Facebook research, and could prove that Facebook has the tools on hand to easily find and remove the content on its platform,” notes Köppen.
Crossing over: the value in interdepartmental collaboration
Köppen’s call for collaborative teamwork stems from the successes she’s had by crossing disciplines and departments, particularly in accountability reporting.
“You have to have a journalistic perspective to really make the right decisions and you have to have a technological understanding and we’re thinking about those persons as tech translators, like people in the newsroom who understand strategy, and we can also see beyond the newsroom, what is going on there.”
“Having those people in the newsroom is key right now. And this is where we have to build up talent pipelines, and very often those people’s people are already in your newsroom, and they’re extremely underestimated. They are sitting somewhere in the social media department, or working as copy editor.
“Those people are really the interesting ones; you want them to make decisions, to consult, and you have to find them within your newsroom.
And once you have them? “I think investigation is what you can do with an interdisciplinary team. You can set up experiments, you can do statistical experiments – and you can also use those investigations to invent new investigation methods. You’re always ping pong in this with people from academia to peer review or journalism to see if our statistical methods are correct.
‘Those are the stories that really drive interest. We can see that people see that as a USP for our news brand, and I can only encourage others to invest in those interdisciplinary teams, and in this beat.’
BR has launched a white paper on accountability reporting.
The post Silos no more: ‘AI demands cross-functional collaboration’ – Uli Köppen appeared first on WAN-IFRA.