Mid-June is always like Christmas and Easter coming together – it’s time for the Digital News Report, our annual survey of what people do with and around news in now 48 countries. It is, to my knowledge, the largest world-wide audience survey of its kind.
My colleagues, led by Jim Egan, have been hard at work and you should definitely read the the full report here and explore the data yourself here. The below, is my own quick summary of their findings on AI, sprinkled with bit of my own thinking. My colleague Amy Ross Arguedas has written an in-depth piece on the AI findings from the report which you can and should check out here.
So what are the main DNR findings?
AI use for news is (predictably) up again
The use of AI chatbots for news is growing quickly but not as quickly as AI use for other purposes: 10% of people use AI chatbots for news, up from 7% last year. However, only 1% say AI is their main source of news, suggesting it currently plays a complementary role for most users.
Who uses AI chatbots for news? Usage is predominantly by those most interested in news and those who trust AI chatbots more, and is more concentrated among younger audiences (16% of under-35s report using AI chatbots for news). South Korea, Greece, and Spain all saw use of AI chatbots for news double year-on-year, but growth was not universal. Several markets, including the USA, the UK, France, and Germany, reported no increase in AI chatbot use for news over the year. In general, the adoption of AI chatbots for news is higher in countries where search, social, aggregator, and video platform adoption is relatively high, and lower in countries where direct access remains popular.
Looking at how people use AI chatbots for news, users seem to appreciate the extra depth or explanation which comes with using chatbots for news; the most popular feature (42% of respondents) is the ability to ask follow-up questions, but also popular are getting the latest news (35%), summarisation (34%), and making news easier to understand (30%). Using AI to evaluate news sources ranks highly where press freedom is under threat or trust in news media is particularly low. And contrary to some expectations, we do not really see people converting content between formats, perhaps because this does not yet work very well. Looking at why people use them, the answer is above all depth and convenience. Globally, the most cited motivation is wanting more depth or explanation (42%), with 39% saying that AI is a faster way of getting news.
However, trust in news answers from AI chatbots is low, sitting at 20% globally (relative to trust in news overall at 37%). In the UK the corresponding figure is 6%, also the lowest of any market.
A recurrent question is: are people clicking through to original sources when they use AI chatbots for news? The click-through data is something we should treat with a lot of caution, as here things like recall bias and stated versus actual behaviour will kick in hard. Overall, 42% of AI chatbot users for news say they always or often click through from chatbot answers to original news sources, but there are wide variations by market — and if we put it in comparison with other platforms and pathways to news (e.g. search engines and social or video networks), we see people are less likely to click through on AI chatbots than elsewhere. And why are people clicking through if they do? The data suggests that AI chatbot users appear somewhat less driven by a need for detail and more by a desire to verify information or understand it.
Finally, we also see a good increase in worries about what is real and fake on the internet, with the concern about fake news having increased by 4pp to 62% – most likely, because people realise that AI means a “difference that makes a difference” when it comes to the production of false or misleading information, as my colleagues and I argue in a recent pre-print (see below).
What should we make of all this? To some extent, it depends on who you look at. AI chatbots will likely become a more significant part of news use over time. For publishers I think the answer is that things will not get easier. Audiences might not use them that much now, but I think this is bound to change and raises questions about what value they find in what news producers but also others have to offer. For AI companies a lot of this is certainly good news and speaks to the usefulness of the systems they have built and deployed for many people. But it also shows they have clear responsibilities, not just towards those producing and providing the content these systems draw from to work – but also towards users as these increasingly use AI to inform themselves (besides many other things they do) apart daily affairs and the world they live in.