Newsletter

Reuters Institute offers insights and a warning in the use of AI and journalism

The public’s attitude about AI is still evolving, but it could have a negative impact on reader revenue, a study has found.

An illustration of a robotic typewriter. AI is challenging media leaders to rethink their business models and focus on where they add value in this rapidly shifting world. Illustration created with Midjourney
AI is challenging media leaders to rethink their business models and focus on where they add value in this rapidly shifting world. Illustration created with Midjourney

During the recent News Rewired conference in London, one of the speakers jokingly said that it had been 25 minutes since anyone had mentioned AI so he felt compelled to do so. Since the introduction of ChatGPT in November 2022, AI has dominated discussions in media circles. Will it accelerate the extinction of legacy news publishers? Or is it a once-in-a-generation opportunity that publishers must seize?

Public uptake of AI will drive the use of the technology, Rasmus Kleis Nielsen of the Reuter’s Institute has said, and now Rasmus and Dr. Ricahard Fletcher have investigated public attitudes towards generative AI and its application to journalism in six countries – Argentina, Denmark, France, Japan, the UK, and the USA. YouGov polled roughly 2000 people in each country to assemble “nationally representative quotas for age group, gender, region, and political leaning”.

The research found that the hype about AI was far greater than the use of it across the six countries, with awareness of genAI services being lower in the UK than in the other five countries. Moreover, those polled in the UK were also far less likely to trust journalists in using AI in their work. The study also highlighted ways people expect journalists to responsibly use the technology, which can help inform publishers’ guidelines on how to use genAI.

The research also offers a warning. The public can see how genAI helps publishers, mostly in cutting costs, but they still aren’t sold on how it helps them in terms of providing more relevant and trustworthy news.


As well as the attitudes about AI, they wanted to answer this question:

(T)here is also a fundamental question of whether and how the public at large will react to the development of this family of products. Will it be like blockchain, virtual reality, and Web3? All promoted with much bombast but little popular uptake so far. Or will it be more like the internet, search, and social media – hyped, yes, but also quickly becoming part of billions of people’s everyday media use.

Richard Fletcher and Rasmus Kleis Nielsen

When your head is spinning inside the media bubble, it is sometimes difficult to know whether the topic du jour resonates outside of it so I popped AI, VR, crypto and blockchain into Google Trends to see how frequently the terms were searched for worldwide.

AI is in a class by itself. VR and crypto/blockchain might have dominated the conversation in tech circles. Still, they are minority sports in terms of public mindshare, and over time AI has had a higher level of interest than these other technologies.

Awareness outpaces adoption

But where does AI stand now? While hardly surprising, more people are aware of genAI than are currently using it, but it is a surprisingly wide gulf, particularly given its widespread coverage.

The introduction of ChatGPT on 30 November 2022, marked a watershed moment for genAI, when a new generation of tools burst into the public consciousness. A year after its introduction, the ‘viral’ app had more than 1.7m users worldwide. It gained 100m users in two months, faster than any other technology introduced in the past 30 years, according to PwC. (It held that crown only briefly because Meta’s introduction of Threads grew faster, racking up 100m users in five days due to leveraging Instagram’s existing user base.) Due to that early lead, Google’s Gemini and Microsoft’s Copilot are playing catchup. The use of ChatGPT is two to three times higher than the genAI tools from the US tech giants.

Researchers asked participants about their awareness of 14 AI tools including Snapchat My AI, LLaMA from Meta, Midjourney and Perplexity.ai. The US and Japan had only 19% of respondents who had not heard of any of the 14 tools. The UK had the least overall awareness of AI tools, with 30% of respondents saying they were unaware of any the tools.

However, the Reuters Insitute research found that genAI has yet to become indispensable to most people’s daily lives. “(F)requent use of ChatGPT is rare, with just 1% using it on a daily basis in Japan, rising to 2% in France and the UK, and 7% in the USA,” the study found. As Dr. Fletcher told the BBC, there is a “‘mismatch’ between the ‘hype’ around AI and the ‘public interest’ in it”.

For journalism leaders, it is important to note that the research found that only 5% of the people surveyed had used genAI to get the latest news, ranging from 10% in the US and only 2% in Denmark and the UK. This is far fewer than the 24% who had used genAI to get information of any sort or to create media, which stood at 28%.

Amongst the countries, educational attainment only had a slight effect on the use of genAI. Age was much more of a factor, with 56% of 18-24 year olds having ever used ChatGPT, while only 16% of 55+ year olds had used the service.

Attitudes towards genAI’s impact

On average across the six countries in the study, 72% of respondents think that genAI will have a large impact on search and social media companies, the largest response in the study. Journalism and science are second, with 66% of those polled saying genAI will impact these activities. Roughly half think the technology will have a large impact on politicians and their parties.

Younger respondents, who the study pointed out are far more likely to have used the tools, were much more likely than older participants to expect genAI to have a large impact in the next five years and that it would make their lives and societies better.

The researchers also asked people how much they trusted various professions, institutions and industries to use genAI responsibly. Healthcare professionals and scientists scored the highest, while respondents were far more sceptical that the news media, national government, social media and politicians would use the technology responsibly. While the US is generally seen as mistrustful of journalism and the government, US participants in the research were much more likely to believe that the news media (30%), the national government (28%) and social media companies (27%) would use genAI responsibly, particularly compared with the UK. Only 12% of those polled in the UK strongly or somewhat trusted the news media with the technology, 13% trusted the national government, 9% social media and only 7% trusted politicians and political parties.

Across the six countries, people were most hopeful that genAI would have a positive impact on scientific research, healthcare, transportation and shopping, and they were most pessimistic about the impact on equality, job security and news journalism. When it comes to media and journalism, the picture is complex. Across the six countries, the difference between those who thought it would have a positive impact versus those who thought it would be negative was +17 when it came to entertainment. Attitudes about the implications for journalism were positive in Argentina (+19) and Japan (+8). Of note, respondents in the UK were particularly negative about its potential impact on journalism, with a net score of -35. That reflects a general mistrust of journalists, “where only one-third of “where only one-third of the population say they ‘trust most news most of the time’”.

Attitudes towards genAI’s use in journalism

A substantial minority think journalists already use AI tools for various tasks including translation, grammar and spelling (guilty as charged) and data analysis. The researchers thought some responses were “a degree of cynicism about journalism from some parts of the public”. For instance, almost a third of those polled thought that journalists used an AI-generated image even when a real photograph was available or 17% believe that journalism organisations use an AI presenter.

The public is much more comfortable with using genAI in covering soft topics like sport and fashion than hard news including crime and politics. However, researchers pointed out that “much of the public does not have strong views either way, at least at this stage”.

However, the research did outline areas where publishers should be concerned about its use. Those polled thought genAI would make news cheaper to make, more up to date and easier to understand while also thinking it would result in news that was less relevant, transparent and trustworthy.

People were more comfortable with journalists using genAI for basic editorial and production tasks such as checking spelling and grammar or translating content into different languages. They were far less comfortable using it to create an image if a real photograph was available or create an artificial presenter.

Essentially our data suggest that the public, at this stage, primarily think that the use of AI in news production will help publishers by cutting costs, but identify few, if any, ways in which they expect it to help them – and several key areas where many expect news made with AI to be worse.

Richard Fletcher and Rasmus Kleis Nielsen

With publishers looking to earn more revenue from subscriptions and membership, they should approach genAI responsibly and ensure they communicate how humans are in the loop of their AI efforts. Across the six countries, respondents were asked “if news produced mostly by AI with some human oversight was made the journalism worth more or less than that which was produced entirely “if news produced mostly by AI with some human oversight is more or less worth paying for than news produced entirely by a human journalist”, 41% said it made it worth paying less for. Only 8% said it would be worth paying more.

And for any content “produced mostly by AI with some human oversight”, only 5% of those polled thought it shouldn’t have a disclaimer. And the researchers said that this might have a negative effect because while people wanted the disclosure, it hurt trust.

Overall, attitudes about genAI are still forming. Young audiences are using it more and are much more comfortable with it. This research offers early insights into where the majority of the audience is comfortable with its use and also clear warning signs about how its use might affect subscription efforts that are key to many publishers’ strategies.

Industry News

Here are some of the most important headlines about the business of news and publishing as well as strategies and tactics in product management, analytics and audience engagement.