How audiences in six countries think of generative AI in news

Earlier this week the Reuters Institute published a new report on how audiences in six countries view the rise of generative AI and its possible impact on society and on the news landscape. The report is authored by our colleagues Richard Fletcher and Rasmus Kleis Nielsen and is based on fresh survey data from Argentina, Denmark, France, Japan, the UK and the US. As you scroll down, you'll find four key findings from the report, which you can read in full in this link .?

1. Many people have heard of ChatGPT, but only a few use it regularly (and even fewer use it for news)

Our data shows that both awareness and use of generative AI are still very limited. A sizable minority, ranging from 19% in the US to 30% in Japan, hasn't heard of any AI chatbots. ChatGPT is the most popular by far, with 50% of the people aware of it. But even ChatGPT is only used regularly by a very small minority of the audiences we surveyed: 1% use it daily in Japan, 2% in France and the UK, and 7% in the US.

The young use AI more. Use of ChatGPT is slightly more common among men and those with higher levels of education, but the biggest differences are by age group. Only 16% of people aged 55 and over say they have used ChatGPT at least once, compared to 56% of people aged 18-24. Up to 27% of people in this younger group use ChatGPT at least once a week.

Only a few use AI for news. Our figures suggest most people use ChatGPT almost as much for creating media (28%) as for getting all kinds of information (24%), from summarising text to getting advice and answer factual questions. Only 5% of the people in our sample say they use ChatGPT to get the latest news. Figures vary from 2% in Denmark and the UK to 10% in the United States.?

2. Audiences prefer news produced by humans, but they are comfortable with certain tasks and topics?

Most people in our survey think that journalists are already using generative AI for certain tasks. For example, 43% think they are using AI for spelling and grammar, 29% for writing headlines and even 27% for writing articles. Perhaps surprisingly, only one-third of the people in our sample think that editors always or often check AI outputs before publication to make sure they are correct.

How audiences think journalists should use it. Audiences are generally more comfortable with news produced by human journalists than with news by AI. However, it's important to stress that younger people are more likely to say they are comfortable with using news produced in whole or in part by AI, and that audiences are more comfortable with AI being used for softer news topics such as sports, fashion and culture.

AI authors and presenters make audiences uncomfortable. Audiences are not equally comfortable with journalists using AI for different tasks. As the chart below shows, across the countries covered, most people are comfortable with using generative AI fr spelling and grammar (net score of +38 points), translations (+35) and creative charts (+28) and very uncomfortable with using it for recreating images (-13) and creating AI authors and AI presenters (-24).

3. Many think AI will make news less trustworthy, cheaper to make and less worth paying for?

Between one-third and half of our respondents do not have a strong view on how generative AI will change the news. Amongst the rest, most respondents expect news produced mostly by AI to be less trustworthy (net score of -17 points) and less transparent (-8). They also expect it more up to date (+22) and cheaper to make (+33). With the exception of Argentina, the balance is the same for these attributes.

Worth paying for? An average of 41% across the 6 countries covered say that news produced mostly produced by AI will be less worth paying for than news mostly produced by human journalists. Only 8% say they think that news made in this way will be more valuable. Even younger audiences share this view. Percentages are a bit higher in Argentina (15%) and the United States (14%).?

4. Many people expect a big impact on journalism and fear news media use AI irresponsibly??

Most people expect generative AI to have a large impact on every sector of society in the next five years, from political parties (51%) to science (66%) and news media (66%). However, while half of the people in our sample trust scientists will use AI responsibly, less than one-third trust politicians, news media and social media companies to do the same.

Charting distrust vs impact. As the chart below shows, there are sectors where fewer people expect a large impact, and more people are worried about irresponsible use (e.g. government and political parties) and also sectors where more people expect large impact, and more people fear irresponsible use by the actors involved (e.g. social media and the news media). As the chart also shows, most people trust scientists, healthcare professionals and retailers to use AI responsibly.?

You'll find many more findings in the report, which is supported by seed funding from Reuters News and made possible by core funding from the Thomson Reuters Foundation. You can read the report in full in this link . You can explore our work on AI and the future of news in this link .