A.I. adoption in media
Conversations with Data: #56
Do you want to receive Conversations with Data? Subscribe
From personalisation to content moderation, artificial intelligence touches many facets of journalism. Still, many hurdles exist for A.I. adoption in the newsroom. Among them include a lack of funding for media innovation and severe knowledge gaps around these A.I. technologies.
In this week's Conversations with Data podcast, we spoke with Professor Charlie Beckett, the founding director of Polis, the LSE's media think-tank for research and debate around international journalism. He talked to us about Polis' JournalismAI project and the findings from its global survey of journalism and artificial intelligence.
External content from a social media channel that would like to place or read cookies is embedded here, for which you have not given permission. click here to change your preferences
What we asked
Tell us about your journey from journalism to academia.
I was an incredibly traditional journalist back in the last century. I started on local papers and worked my way up through television news at the BBC and so on. Back in 2006, I left Channel 4 News to become director of a new journalism think tank at the LSE called Polis. It was a bit like doing a start-up as I was given this blank sheet of paper to create it. At the time journalism was exploding with all these new technological waves -- from the move online to the explosion of social media. I wrote a book about network journalism back in 2008. My work has been about exploring all these digital technologies and understanding their impact -- all the stuff I couldn't do when I was a journalist. And for me, that was the attraction of A.I., the latest wave of this technologically driven change.
As a media think tank, what is Polis' mission?
Polis is a journalism think tank within the Department of Media and Communications at the LSE. I spend all my time working with journalists or with people who are affected by journalism. It aims to provide a forum to talk about journalism and its impact on society and to do research, of course. But primarily, it is to be a bridge between the researchers, the academics and the journalists themselves. And increasingly, it's become a bridge within journalism. Polis' JournalismAI project is very much about connecting journalists around the world in small and large newsrooms as well as different sectors. We share knowledge and best practice.
How did Polis' A.I. journalism work emerge?
Last year we received a subsidy from the lovely people at Google, which allowed us to do this global survey. The first thing was to find out what was happening. The survey pointed to an incredibly varied and rich picture of all sorts of activity in all sorts of newsrooms in all or parts of the news process -- all the way from newsgathering to news delivery. The survey also told us about where the gaps, the challenges and the problems were for A.I. adoption and adaption.
Now we are in the second year, where we have been trying to find not solutions, but at least pathways. People told us was that there was a huge knowledge problem as well as an understanding problem. That led us to create a free training course in collaboration with VRT News and supported by Google News Initiative. People also told us that they were really struggling to find the time and resource to do innovation and to prototype new ideas. This led us to create an experimental collaboration project to do precisely that.
Tell us more about this experimental collaboration project.
It's called the JournalismAI project. We've got more than 40 journalists from every kind of news organisation around the world, from Hong Kong to Argentina. They've divided into groups and are working on a series of challenges. Those could range from how do we get more editorial diversity in our content to how do you maximise the revenue from subscription models. They've just been working for a couple of months, and their energy and their thoughtfulness has been inspiring. They're sparking off each other, sharing experiences, coming up with ideas that they wouldn't necessarily be able to do in their own newsroom. We've got a group of fantastic coaches who are editorial innovators to help out too. The idea is that by the autumn, they will come up with something. It probably won't be a finished, beautifully produced tool or system. Still, they'll come up with a prototype or an idea for a prototype or system that could be useful in many news organisations.
Where should a journalist start if they want to learn more?
Start with our one-hour training course, Introduction to Machine Learning. It is built by journalists, for journalists, and it will help answer questions such as: What is machine learning? How do you train a machine learning model? What can journalists and news organisations do with it, and why is it important to use it responsibly? There are some great books out there, too. Nicholas Diakopoulos' book gives a more academic take on automating the news. There's also Francesco Marconi, who's written a much more newsroom focussed A.I. in journalism book. People are starting to create courses, but they are more academic at this stage. Journalism schools are still catching up.
Lastly, what is the biggest hurdle to A.I. adoption in the newsroom?
Money. It's not just money in terms of let's spend the cash to get the shiny new things. It's investing the time and money in the long term. It can take you three to six months to develop it followed by another six months to iterate it. And then you still have to pay attention to it. I think that is the biggest problem is applying those resources and having the general knowledge in the newsroom to do that efficiently. There is no easy answer to that, I'm afraid. We hope that as these technologies become more mature and more adapted, that they'll be replicable.
Machine learning has been around for a while, but it is rapidly developing in almost every field. But the difference is that pharmaceuticals or fintech have got gazillions of dollars to develop these and it can do it at scale. The news industry is relatively small and has far less money for independent R&D. And that's why -- love them or loathe them -- we should work with the tech companies, universities and other partners who can help news organisations to see what might work for them.
Latest from DataJournalism.com
As Black Lives Matter protests spread around the world, journalist Sherry Ricchiardi examines the verification methods and OSINT tools journalists and non-profits like The New York Times and Amnesty International used to portray the movement in the wake of George Floyd's death. The long read article also points to useful resources from the Global Investigative Journalism Network, First Draft News and the Committee to Protect Journalists. Read the article here.
Our next conversation
The Conversations with Data podcast will be back in September. We will discuss social network analysis -- an advanced form of analytics that is specifically focused on identifying and forecasting connections, relationships, and influence among individuals and groups.
As always, don’t forget to let us know what you’d like us to feature in our future editions. You can also read all of our past editions here.
Tara from the EJC Data team,
P.S. Are you interested in supporting this newsletter as well? Get in touch to discuss sponsorship opportunities.