The “tsunami of AI content” could lead audiences back to trusted journalism sources
Readers First Initiative Blog | 16 July 2024
Leading up to the INMA Latin-American Conference this week, I sat down with Gastón Roitberg, editorial secretary of Argentina’s La Nacion, to discuss content, technology, and the news media business.
I thought you’d be interested in the content of the interview and have republished it here. I look forward to speaking at the conference on Friday. Are you attending? More information here.
How can traditional media adapt to compete with emerging digital models?
One obvious approach is to follow consumers: Publish on the platform, channel and format of their choice. It doesn’t necessarily mean that all news outlets like La Nacion should publish short-form entertainment videos on TikTok. Different people have different preferences.
I would recommend viewing the audience not as a whole but as a series of segments. There are audiences around the world who still prefer to read a desktop Web site or even a digital replica of a print newspaper, who visit their favourite news Web sites directly and who prefer to read articles, and some even read longer ones. These audiences tend to be older, better educated, wealthier, more politically active, and willing to pay more for quality journalism.
These are the target audiences for paywalls. As Big Tech companies have taken over the lion’s share of advertising revenue, many news publishers are adapting their business model and looking for ways to increase reader revenue as a primary alternative to advertising.
What impact have you seen of Artificial Intelligence (AI) in news production?
AI can generate text, images, speech, music, video, but it also plagiarises other people’s work, makes stupid mistakes, and confabulates. It is therefore unreliable as a substitute for human journalists or editors in the news value chain.
In general, AI is unlikely to steal entire jobs from humans, as jobs are a set of tasks: in addition to writing, a journalist witnesses events, researches and verifies facts, makes sense of them, and offers an interpretation. None of these can be reliably automated. And the higher the stakes, the higher the risks.
At the same time, some individual tasks, such as proofreading, translating, or reading texts aloud, can be automated with AI, and we see rapid adoption of tools that perform such tasks around the world. This is already affecting people’s work.
A recent study by researchers at Harvard, TU Berlin, and Imperial College estimated the impact on freelance job postings since the emergence of ChatGPT and Midjourney: The number of postings for journalists has dropped by 21%, and the number of postings for text-related tasks has dropped by 17%.
So instead of hiring a freelancer, people hire an AI platform. According to INMA surveys, news publishers around the world are already using AI to come up with headline test texts, summarise or tag articles, write posts for social media, and so on.
Some publishers are going even further: creating synthetic TV presenters to produce hourly weather shows, adding audio articles using their authors’ voices, and launching products in new languages.
In short: AI can lead to gains in efficiency, productivity, and in effectiveness in news production and distribution.
In this AI advancement, how can we generate value for the audience if we don’t need to intervene?
We are facing a tsunami of AI content.
NewsGuard — a US non-profit organisation — has identified hundreds of Web sites publishing numerous AI-generated news articles every day. These AI publishers are flooding the market with cheap content, and in the short term, AI will likely lead to a decline in the overall quality of the Web and misinformation.
In the long term, I am optimistic. The AI tsunami could soon lead to audiences becoming frustrated by low-quality content and valuing quality journalism.
How can the media effectively combat disinformation and “fake news”?
The media are the good guys here. They should focus on what they do best: producing accurate, fact-based, impartial and responsible journalism.
At the same time, governments and citizens should defend themselves against the pollution of public space with disinformation and “fake news,” in a similar way to how they defend themselves against air or water pollution.
Companies that enable polluters (in the case of news: Big Tech platforms such as Google and Meta) should be held accountable and responsible for reducing harm to society. For example, they should be required to actively limit the distribution of disinformation, support news literacy education, and contribute fairly to the funding of quality journalism, for example through taxes, licences, or other collective agreements.
What do you think is the role of paywalls and subscriptions in the digital media business model?
I am glad to see a widespread willingness to pay. According to the latest Oxford’s Reuters Institute’s Digital News Report, across 20 countries, 17% of consumers are already paying for or using paid news online. Furthermore, 36% said they would consider paying if the price was right.
This means that in total 53% of consumers internationally are willing to pay for online news. That is an opportunity to triple current penetration. In terms of revenue mix, consumers will already have funded most journalism worldwide by 2023, according to WAN-IFRA World Press Trends.
By the way, consumers also funded most media overall — of the total US$2 trillion global Internet and media market, one-third was advertising revenue, another third was paid content, and the rest was the cost of Internet access paid by consumers.
In my view, reader revenue is well aligned with journalism, perhaps better than advertising. Why? It motivates media outlets to focus and innovate around the needs of audiences rather than governments or advertisers. It incentivizes quality and uniqueness of content. It provides the predictability and stability needed to invest in journalistic talent and skills. It leads to fewer ethical conflicts than other businesses, such as advertising, oligarch sponsorship, or government subsidies.
Still, advertising and other revenue streams play a role in the future of the news business. From our research, we know that diversification strengthens the profitability and stability of media companies and helps reach broader audiences.
What are the most effective strategies to increase audience engagement on platforms?
The most valuable customers are those who visit regularly: They are more likely to pay for online news and see the most ads. So the most effective engagement tactics focus on helping people develop the habit of checking the news regularly.
Examples? Media companies update news on their home pages to attract direct visitors to Web sites and apps. They send newsletters via e-mail, mobile push notifications, or text messages. They publish digital replicas of print newspapers for older readers. They launch podcasts. They add crossword puzzles and other games, or cooking recipes.
Some publishers also see value in building communities around their brands and their journalism. Encouraging peer-to-peer discussions and sharing creates a sense of belonging and loyalty. Studies also show that a commentator views many pages before and after posting a comment, so they see more ads and are stopped sooner by paywalls.
What challenges do local media face compared to national and international media?
Local newspapers were traditionally very profitable because they often enjoyed a local monopoly on the production and distribution of advertising in their geographic area.
This privileged position has disappeared, and the majority of advertising spending goes to Google and Meta.
This new market reality requires local media to become much more efficient than in the past and work with lower margins than monopolies. At the same time, local media must invest in digital transformation: adopting new technologies, attracting talent, and developing data skills.
How to afford such investments when money is scarce? This conundrum has led to the consolidation of the local news industry and the emergence of super-regional groups. It is easier to spread the cost of technology across a larger number of brands and a larger group of customers. Across the world, we see greater profitability from national publishers or local publishing groups than from individual local media companies.
It is therefore crucial to use the available data to improve content and distribution strategies?
Data is basically a kind of consumer feedback. It helps news editors make better decisions about what journalism to produce and how to deliver it.
Knowing audiences helps newsrooms understand their needs, focus their work on meeting them, and helps motivate people to do what matters. In this way, newsrooms can outperform others that don’t know consumers as deeply.
Data on people’s behaviour also helps us group them by interests or behaviours and then tailor their experience with journalism accordingly. Do you like football? Why don’t we feature this story at the top of your home page?
Of course, data analysis has limitations: It can only tell you about what happened in the past. You can extrapolate data and try to predict future behaviours, but it won’t tell you anything about what might happen but didn’t happen. Data won’t tell you what’s right or wrong, either.
In quality news organisations, data informs decisions, but it doesn’t drive them. Editorial judgment remains critical in journalism.
What role does content personalisation play in audience retention?
In general, we personalise because we expect it to increase relevance and therefore drive usage and satisfaction.
Personalisation means different things to audiences: Some expect a personalised selection of articles, others want articles tailored to their circumstances. There are also people who seek a personal connection with authors.
Technology allows us to personalise almost everything, from ads to offers and subscription messages, landing pages, newsletters or recommendation feeds, and articles. Personalisation is proven to increase click-through rates for specific offers or recommendations, and this leads to higher conversions and engagement.
What technological innovations do you think will be crucial for the future of journalism?
There are many technologies that promise to reinvent media: Generative AI for one-to-one personalisation, Augmented Reality for immersive storytelling, blockchain for content verification.
At a time when so much is changing, it can be useful to stop and think about what actually stays the same. The core mission of journalism, which is to seek truth and hold power to account, remains unchanged. Storytelling is a fundamental human need, and while channels or formats change, a good story is always a winner. Ethics and trust remain vital in a world of AI content tsunamis and deepfakes.
How do you see the balance between the need for quality content and the pressure to generate clicks and visits?
Balancing volume and value is one of the most difficult tasks in news media management, and everyone struggles with it, from The New York Times to Long Island’s local Newsday.
First, it helps to be clear about who your primary customer is (a reader or an advertiser) and what your North Star goal is. Clarity in goals helps you set the metrics that matter and optimise them.
Second, it helps to think of your content as a portfolio. We want to maximise stories that accomplish both: deliver quality and generate clicks. Some stories will excel in quality but not in clicks, and we’ll still run them because we care. Some stories will be less sophisticated but will generate lots of clicks, and we’ll do them to meet our business needs. We can stop doing stories that don’t deliver quality or attract clicks.
Third, some publishers try to automate these decisions and, for example, optimise links on home pages or social media posts for different goals in real time. A human still designs those algorithms, so human judgment is still key.
What skills do you consider essential for journalists who want to excel in an ever-changing environment?
If you look at job postings, you might think that the essential skills today are data literacy and multimedia storytelling. I think these can be taught in schools or trained on the job. What is really essential is curiosity, listening skills, and honesty.
In the coming years we will see a struggle for journalistic relevance?
Many believe that news outlets’ relevance depends on doing what the audience seems to want: short-form entertainment videos on Tik Tok.
Studies on willingness to pay for journalism find something else. When it comes to value, people don’t care much about videos or social media posts.
Researchers at City University London and Ludwig Maximilian University of Munich found that willingness to pay was higher when audiences were informed that their subscription would support independent, inclusive, and power-checking journalism, and that a subscription model was implemented due to the critical financial situation of the news industry.
So how do you strive for journalistic relevance? Stay true to your mission, do what matters, and be honest with your audience.
Greg’s Readers First newsletter is a public face of a revenue and media subscriptions initiative by INMA, outlined here. INMA members can subscribe here.