“These new technologies are the new printing press”
“Machine learning, automation, personalisation, data analysis and natural language processing tools can supercharge the modern news media,” says Charlie Beckett, Head of the Global JournalismAI project. They are all a part of artificial intelligence. However, being a relatively new set of technologies they come with fresh challenges for publishers.
“Reality of the vast majority of media”
A new report by International Media Support (IMS), in collaboration with The Fix and El Clip, looks into how 44 publishers across 20 countries in Latin America (LatAm) and Central and Eastern Europe (CEE) are using AI.
“The current picture of how newsrooms are adopting (or building) AI/ML is a bit one-sided,” comments Jakub Parusinski, Editor, The Fix Media. “Indeed, much of research on the topic to date focuses on larger, wealthier Western newsrooms – who are often on the forefront of innovation.
“This ignores the reality of the vast majority of media from emerging and frontier markets.”
That’s the gap, the report, “The next wave of disruption: Emerging market media use of artificial intelligence and machine learning,” seeks to address.
One of the keys to success with AI is collaboration, says Professor Beckett. “By collaboration I mean working across the usual departmental barriers within the news organisation,” he writes in a foreword to the report. “AI should not be left to the technology team and in turn, the technologists need to be integrated into the news production process.”
Moreover, AI can solve problems across departments like engineering, design, sales and editorial, further strengthening the need for inter-departmental collaboration.
User experience in product design requires cross-functional ways of working. Products need to be designed by cross-functional teams so that they meet all requirements and use all the insights.The next wave of disruption
Publishers should not stop there. Beckett encourages them to build working relationships with other news organisations, technology companies, start-ups and AI labs within universities. “There is so much to learn from other people’s knowledge and experience,” he explains. “It is tough to do it on your own. Other people will have made mistakes or found clever solutions. Learn from them. In the end everyone benefits.”
Collaborative approaches between media, research institutions, or third party solution providers should be encouraged as many applications or ventures are beyond the scope of a single outlet, due to the resources needed, availability of data, or other barriers to entry.The next wave of disruption
“Lead the way with experimentation”
Beckett recommends starting with educating people in the organization about AI for journalism. This would include fighting fears of job-loss and pointing out new possibilities. It may require fresh recruitment or collaboration with experts as well.
Publishers should not be deterred by fears that implementing AI could be expensive or require considerable resources, the authors suggest. “Larger media may deploy a broader range of solutions,” they explain, “but small outlets can build and use a range of tools and lead the way with experimentation.”
AI bridge roles can be very effective in setting the wheels in motion. “The presence of translators or experts that others can reach out to for all artificial intelligence matters, appears to be crucial to the success of subsequent implementation of AI/ML use cases.”
The next step is forming an interdisciplinary team to build foundational blocks for future projects. It can begin with simple ones like indexing archives or connecting systems through APIs. The findings show that there are opportunities to leverage automation wherever there are repetitive, process-heavy tasks. They can be found across an organisation from editorial to internal operations and product development, to commercial functions.
“Roughly 40% of AI/ML use cases are in editorial, 40% goes to commercial, and 20% covers internal functions,” says Michal Cyrek, Big Data Architect at Onet.pl, one of Poland’s biggest digital media.
“Integrating AI into the workflow is not rocket science”
“Start with a project you know you can handle that will deliver an immediate impact,” the authors suggest. “It will help set the tone and make people realise integrating AI into the workflow is not rocket science. You will also start adding valuable know-how into the company that you can later use for more complex projects.”
Your first project is all about getting people on board. A quick win will boost confidence and results will inspire people to tackle new problems.The next wave of disruption
Going beyond their own organisations, publishers should look at collaborating with other media outlets to solve similar problems together. They can also partner with tech companies, use international workshops and resources to introduce new technologies into the newsroom.
Most publishers don’t need to/ may not have the resources to develop in-house solutions. The report encourages them to use vendors or open-source solutions to implement AI faster. The authors recommend choosing solutions that can be measured against KPIs. This will make it easy for everyone to see the benefits.
We always start from the user. Whatever raises time on site, frequency of return – anything for a sweet and smooth UX.Michal Cyrek, Big Data Architect, Onet.pl
For example, natural language generation (NLG) can be used to automate content creation. It’s useful for covering topics that require both public and private structured data sets like sports, real estate or company news. The authors underline that using NLG does not mean the journalists creating such articles would lose their jobs. On the contrary, they will have more free time to create higher quality content. They can even creatively enrich the automated copy. The publisher on the other hand will be able to dramatically increase content creation.
Infobae, an Argentinean news website, boosted automated content production 10x since 2019. It expanded to 15 verticals without increasing headcount.
Print subscriptions up 80% and digital doubled using algorithms
The report finds that management of paywalls and subscriptions are the most widely used AI applications in both LatAm and CEE. Colombia based El Tiempo has grown its print [subscriptions] by 80% and digital by 2x using algorithms, according to its Data Strategy Manager, David Rodriguez.
Some of the popular third-party paywall management solutions available to publishers include those by Piano and Deep BI. Dennik N in Bratislava built its own, open-source platform called REMP2020 which has been upgraded to REMP2030. The upgrade introduces machine learning solutions from the ecommerce industry. Dynamic paywall, user behaviour/churn prediction, content evaluation powered by AI/ML are some of the common features offered by these products.
It doesn’t make sense to try to develop something by ourselves. Especially when there are so many simple solutions out there.Konstantins Kuzikovs, CEO, Delfi Latvia
“Third wave of disruption transforming news media fabrics”
“There is still a huge gulf between the possibilities that exist in the use of technologies (AI/ML) and what is being done in the newsrooms,” says Ana Laura Perez, Digital Product Manager, El País, Uruguay. “If writers, editors and managers knew how to start incorporating them, it facilitates the quality of the work, it would completely change the final product and the need to invest in these types of tools would become more evident.”
The single biggest challenge for media to adopt AI-powered solutions, according to Jaroslaw Gora, CEO, Deep BI, is finding people within newsroom leadership who understand the basic principles of their solution, and can integrate it into their workflows. It applies both ways – whether a publisher chooses to use third-party solutions, or develop them in-house.
“Technical aspects are not the hard things, it’s always the culture,” adds Erik Van Heeswijk, CEO of AI-driven editorial analytics firm, Smartocto. “Everybody needs to be involved.”
“Our research on AI and journalism around the world showed that there are increasing inequalities for news organisations,” adds Beckett. “The danger is that local news and media in emerging economies might fall behind.”
The big global brands have the R&D resources to buy in or develop their own systems. But the others will have to work together if they want to capture the benefits of AI for themselves.Charlie Beckett, Head of the Global Journalism AI project
“Artificial Intelligence is the third wave of disruption transforming news media fabrics,” the authors write. The first two being going online and the advent of social media. “It offers an opportunity for the media to make significant steps around content production and data.”
These new technologies are the new printing press.María Florencia Coelho, New Media Research Manager, La Nación in Argentina
The full report can be downloaded from IMS:
The next wave of disruption