When we talk about journalism and artificial intelligence (AI), we often assume that we just bring a piece of tech to the newsroom to fix an editorial problem and carry on doing journalism as usual. But what makes us so sure that the way we tell stories works in the first place?
How we define a problem has a lot to do with the solution we come up with. While most news organisations are looking at using technology to grow their audience and get their attention, few are wondering whether the kind of journalism they are using the tech for is actually useful to their readers.
That is exactly the question that Pierpaolo Bozzano, head of Content Innovation Lab at the Italian publisher Il Sole 24 Ore, wanted to explore. So he and his team joined Collab Challenges, a series of experiments organised by JournalismAI, a project of Polis, the journalism think tank at the London School of Economics and Political Science (LSE) and supported by Google News Initiative.
Bozzano started by spelling out the uncomfortable truth about today’s journalism: long articles do not reach most audiences. People just scan the headline and the first few paragraphs, and the rest of the piece sits on the website unread.
So he and his team looked at specific parts of an article to see whether they resonate with the reader. They used AI to tag these parts and realised that there are bits that are genuinely useful and others that are dysfunctional. That means that some of the sentences, quotes and paragraphs we routinely use in journalism are not only pointless but they put people off.
For a lot of journalists and editors, this can be hard to swallow. Bozzano said that there was some resistance from his colleagues who felt like their freedom to build a story the way they want is being taken away. But if you are serious about writing for your readers and not for yourself (or to impress other journalists), you may want to explore a new way to do journalism.
Pick 'n' mix
The concept of looking at a story not as a monolith but as a collection of bits that are useful or useless to the reader is called modular journalism. A module is a piece of text that, for example, answers a specific question like "why is this story important" or "what is the impact on my community". To define the modules - and to teach the tech to recognise them - the team worked with linguists and also took into consideration user needs.
Once you identify the modules, you can then pick and mix them according to who the story is for. This is one of the most important points of modular journalism - we need to stop thinking of our work as one-size-fits-all. The same story can be better understood if it is presented in a different way. For instance, if a user wants a short, bullet-point account and we present them with a 2,000-word feature, we are not serving them properly. The same is true about a story angle - some readers may be more receptive to an opinion while for others, seeing an opinion thrown into the mix in a story that is supposed to inform will make them stop reading.
Bozzano gives an example of financial news stories, which are less often read by women and young people. The modular journalism approach revealed that the problem was not the topic but the way these stories were told.
"We realised we need to stop mansplaining in our articles," he says. Writers help determine how the modules are assembled, but the tech also works to fit them together so they are tailored to the user. If you want to keep women readers' attention, omit the bits that may seem alienating or condescending and use those that bring genuine value.
Journalists are often taught to write in a specific structure and order so the succession of information makes sense. Surprisingly, the story still makes sense when you rejig the modules because these are standalone ideas that make sense in their own right. Take a look at this article by Shirish Kulkarni that has five versions, depending on the target audience.
The modular journalism project is a break from the past in the way journalists address their readers. It is an opportunity for us to step back and stop considering ourselves an elite who knows best what people need.
"It’s humbling and interesting to take interest in the user," says Bozzano. "Trust and engagement are low, we need to do something about it."
In his view, AI is the future of journalism as it can help newsrooms build algorithms to better serve their audiences. On the flipside, it can also give people new tools to control the way they want to consume content. Publishers like the BBC or The Guardian are already exploring new ways to produce content as bits that can be stitched together, rather than feeding everything into a CMS that produces a monolithic article. Ultimately, the prize is user attention and that is the highest currency in the digital age.