GenAI-Generated News: Less Accurate, Timelier, Survey Finds

(iQoncept/Shutterstock)

People are less likely to believe news written by generative AI, according to a recent public opinion survey conducted by Reuters, but the news would be timelier and cost less for the publisher to produce. The online survey also found GenAI tools are not being as widely adopted by consumers as some people may think.

The Reuters Institute for the Study of Journalism (RISJ) at the University of Oxford contracted with YouGov to answer questions in two main areas: Determine how people are adopting GenAI broadly, and gauge their feelings towards GenAI-generated journalism specifically. The survey was conducted online in March and April, with 12,000 responses across Argentina, Denmark, France, Japan, the UK, and the US.

In their report on the findings, Richard Fletcher and Rasmus Kleis Nielsen note that more than half of the survey respondents in the six countries have heard of ChatGPT, making it the most widely recognized GenAI tool.

“That being said, frequent 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,” Fletcher and Neilsen write. “Many of those who say they have used generative AI have used it just once or twice, and it is yet to become part of people’s routine internet use.”

There’s a large chunk of people–19% to 30% in each of the six countries–who have never heard of ChatGPT or any other GenAI tool for that matter. “While many have tried using various of them, only a very small minority are, at this stage, frequent users,” Fletcher and Nielson write.

The percentage of survey respondents who have heard of a GenAI tool in each country (Image courtesy RISJ)

Who’s using it? Predominantly young people, according to the survey, which found 56% of folks in the 18 to 24 year-old bracket say they have used ChatGPT at least once, compared to 16% of those aged 55 and over report having used it once or more.

About one-quarter (24%) of respondents across the six countries say they have used GenAI to get information, and slightly more (28%) say they have used it to generate media, including text, audio, code, images, and video. However, only 5% have used it to get the latest news, the RISJ says.

There is broad consensus among the survey-takers that GenAI will prove to be beneficial in areas like scientific research, healthcare, transportation, shopping, entertainment, education, food and nutrition, and climate change and sustainability. Areas where people have the most concern about GenAI impacts revolve around job security, news and journalism, equality, and cost of living, the survey found.

Most journalists are already using GenAI, according to RISJ’s opinion survey. On average, the survey-takers say that 43% of journalists are using GenAI to edit the spelling and grammar of an article and to translate stories into different languages, and 40% are using it for data analysis. The least likely GenAI tasks, the survey found, include creating an artificial presenter or author (an average of 17% believe this), writing the text of an article (27% believe this) and rewriting the same article for different people (28% believe this).

“In general, the order of the tasks in Figure 15 reflects the fact that people – perhaps correctly – believe that journalists are more likely to employ AI for behind-the-scenes work like spellchecking and translation than they are for more audience-facing outputs,” Fletcher and Nielsen write. “This may be because people understand that some tasks carry a greater reputational risk for journalists, and/or that the technology is simply better at some things than others.”

You can access the complete survey results here.

Related Items:

Is ChatGPT Getting Dumber?

ChatGPT Overtakes Netflix in Web Traffic

OpenAI’s New GPT-3.5 Chatbot Can Rhyme like Snoop Dogg