texas-moody

What do Argentines think of the use of generative AI in the news?

A study on people's perceptions of the use of generative artificial intelligence (AI) in news showed that the most popular platform in Argentina is by far ChatGPT, although very few use it for news. It also showed that Argentines have slightly more confidence than people from other countries that the news media make responsible use of this technology.

The study “What Does the Public in Six Countries Think of Generative AI in News?”, prepared by the Reuters Institute for the Study of Journalism at the University of Oxford, in the United Kingdom, and published on May 28, sought to analyze and document the public’s attitudes towards generative AI, its application in various sectors of society, and, specifically, in journalism and the media.

Argentina was the only Latin American and Spanish-speaking country considered in the research, which also included Denmark, France, Japan, the United Kingdom and the United States. The study was carried out through an online survey of around 2,000 people from each of those countries.

Cover of the Reuters Institute report "What does the public in six countries think of generative AI in news?"

The report shows the results of a survey to people in Argentina, Denmark, France, Japan, the U.K., the U.K. and the U.S. (Photo: Screenshot from the Reuters Institute website)

Although the survey yielded similar results for each country on most questions, some findings regarding Argentina differ slightly from the rest, especially regarding people's level of trust in the use of AI by news media.

“Although some patterns are consistent across the six countries surveyed, our data suggests that people in Argentina are a little more optimistic than those in other parts of the world about the impact that generative AI will have on society and on their own lives,” Richard Fletcher, Director of Research at the Reuters Institute and one of the authors of the study, said to LatAm Journalism Review (LJR). “For example, 44% of those surveyed in Argentina think that generative AI will make society better, compared to 30% on average across the six countries. This is also true of generative AI in news”.

Thirty percent of Argentines surveyed said they somewhat or strongly trust that the news industry makes responsible use of generative AI, while the percentages in the rest of the countries ranged between 12 and 23 percent, except from the United States, whose percentage was also 30.

The results are similar in the field of social networks: 30 percent of Argentines said they trust that social media companies make responsible use of generative AI, while the percentages in the rest of the countries range from 9 to 27 percent.

Similarly, 40 percent of Argentines said they trust that human editors in news media always or frequently monitor the results of generative AI. The figure is also the highest among the six countries and higher than the average of 32 percent.

The report indicates that variation between countries on this issue may be related to differences in levels of trust in institutions in each country. However, although Argentina's figures are slightly higher than those of the rest of the countries, they are still low percentages, so the authors did not rush to draw definitive conclusions.

“A significant part of the public does not have a firm view on whether they trust or distrust different institutions to make responsible use of generative AI. Varying from sector to sector and from country to country, between roughly one-quarter and half of respondents answer ‘neither trust nor distrust’ or ‘don’t know’ when asked,” the report said. “There is much uncertainty and often limited personal experience.”

However, Argentine journalists consulted by LJR consider that the figures found in the study do not necessarily reflect a high level of trust among people in the Argentine media.

Adrián Pino, Argentine data journalist and executive director of the anti-disinformation organization Proyecto Desconfío, said that the study's data could be due instead to the limited knowledge that still exists on the subject in that country and low levels of penetration of that technology in the Argentine media ecosystem.

“Argentina has suffered a great deterioration in trust in the media in recent years. Although this has several causes, the fact that 3 out of 10 users consider that the media responsibly uses AI in Argentina may have to do with the low levels of knowledge that these topics have among news users,” Pino told LJR.

The journalist, whose organization collaborates with small media outlets in Latin America to train their journalists in the use of AI tools for journalistic work, believes that the results of the study could have more to do with the fact that the presence of generative AI is not yet widespread in the Argentine media and people have little reference about its use in journalism.

“Argentina is just taking some interesting steps in the use of AI in journalism. These innovations are not very widespread and still show a low degree of application in daily routines,” he said.

ChatGPT, the most popular

ChatGPT, from the organization OpenAI, is the best-known generative AI platform in Argentina, with 41 percent, followed by Snapchat's My AI (17 percent), Google's Gemini (15) and Microsoft's Copilot (15) , according to the study, which showed similar figures for the rest of the countries.

An average of 24 percent of respondents across the six countries said they had used generative AI to obtain information, while 28 percent said they had used it to generate some type of content, whether text, images, video or sound.

Among respondents who said they use generative AI to get information, only 5 percent on average said they do so to get news. Argentina's percentage for this question is slightly higher than the average, at 6 percent, only below the United States, which got 10 percent.

The report said that a possible reason for the low use of this technology to obtain news is that the free versions of generative AI platforms have been trained with data up to specific dates. ChatGPT version 3.5, for example, uses information through September 2021. Another possible reason, according to the report, is that more and more news outlets are blocking these platforms from accessing their content.

The document indicated that the figure of 10 percent of people using generative AI for news in the United States could be due to the fact that Google's Search Generative Experience (SGE) has been in the testing stage in that country for about one year. The SGE applies generative AI to the company's search engine.

“People who use Google to search for a news-related topic – something that 23% of Americans do each week – may see some generative AI text that attempts to provide an answer,” the report said. “However [...] the higher figure in the USA may also simply be because generative AI is more widely used there generally.”

SGE functions have been available in some Latin American countries since this year, although less widely than in the United States. What does exist in some Latin American countries are applications developed by news media that use AI to offer better navigation of journalistic content, according to Carlos Jornet, president of the Press Freedom Commission of the Inter American Press Association (IAPA), and director of La Voz del Interior of Córdoba, Argentina.

“[The newspaper] Clarín has an artificial intelligence application that makes a sort of summary of the main points of each news item and also highlights figures and phrases. It is a kind of co-pilot for reading the news,” Jornet told LJR. “Perhaps that is what has increased the percentage in some of those consulted [who said they used generative AI to obtain news in Argentina].”

How do journalists use generative AI?

A significant portion of respondents across all six countries believe generative AI will have an impact on communications. Seventy-two percent believe that this technology will affect social media platforms, the sector with the highest percentage in that question. Search engines follow, with 71 percent; and the news media, with 66 percent.

The study also asked whether the impact of generative AI on each sector would be positive or negative. In the six countries, the news and journalism sector ranked on the side of pessimism, only below job security. In this regard, Jornet said that, although AI has the potential to positively impact journalism in Argentina, he also sees great risks in its use.

“I believe that AI well applied in journalism can be a great help, but I also see enormous risks of generating false news, of using it to stigmatize, harass, engage in cyber espionage, even through false information that is being spread by AI robots,” the journalist said. “That is starting to be a serious concern in some countries, so there the role of serious and responsible journalism with traditional brands behind it is essential for people to have confidence.”

More than 40 percent of respondents across all countries believe journalists use generative AI frequently or always for translation (43 percent), spelling and grammar checking (43), and data analysis (40). On the contrary, content generation tasks occupied the lowest levels: generation of headlines (29 percent), images (28) and complete articles (27).

Although the results mostly coincide with how Argentine journalists are using generative AI now, Pino considers that this technology is still in the experimental phase in journalism in his country, and that it will take more time before its use is consolidated in the newsrooms.

“Most cases of media using AI in Argentina are in mechanical tasks, such as transcribing interviews, or the use of tools to identify names or companies in large volumes of documents,” the journalist said. “Another frequent use that we see in various AI solutions is aimed at improving the SEO positioning of news headlines.”

Jornet agrees that the use of generative AI in Argentine news media is not yet systematic, but experimental. In addition, he said that there is a notable disparity in the level of adoption of this technology between large media outlets and small media outlets.

“Perhaps the economic crisis that Argentina, and the media in particular, is experiencing is hindering or delaying this adoption,” Jornet said. “The major media have followed the path of foreign media, which are not only creating laboratories, but are already carrying out concrete applications.”

The journalist said that in the Argentine media there are more notable cases of the use of Natural Language Processing (NLP) models for the analysis of large volumes of data rather than for generative tasks.

Such is the case of “Genie,” the tool developed by the newspaper La Nación, which measures the level of gender representation in the media's content by detecting the gender of the people mentioned or cited in the content. The tool, which has been used internally by the La Nación newsroom since 2023, won the WAN-IFRA Digital Media Awards Americas for Best Use of Artificial Intelligence in a Newsroom.

Jornet also mentioned a case from the newspaper he directs, La Voz del Interior. It analyzed a bill of more than 600 articles presented by the government of President Javier Milei in January of this year. The media outlet, with the support of a communication consultancy, applied AI tools to analyze the meaning and possible consequences of each article of the bill.

“In a very short time we were able to do a point-by-point analysis of that law and a necessity and urgency decree that had also been approved in those days,” Jornet said.

Will AI help media sustainability?

The study asked people how they think news produced mostly by AI with some human supervision will compare to news produced entirely by humans, considering a list of qualities and attributes. On average, respondents across all six countries predict that news produced largely by AI will be less reliable and less transparent, but more up-to-date and – by a wide margin – cheaper to produce.

“Right now audiences can see how generative AI benefits news organizations, but find it harder to see how it might benefit them. This makes it important for news organizations to communicate how their use of generative AI improves and adds value to their coverage,” Fletcher said.

The survey asked whether it would be worth paying for news produced mostly by AI. Forty-one percent of respondents across the six countries said AI-generated news content is less worth paying for, while only 8 percent think AI-generated news will be more valuable.

However, Argentina had the highest percentage of people who believe AI-produced news is more worth paying for, at 15 percent. Thirty-seven percent of Argentines believe that the value of paying for news produced mostly by AI with some human oversight is the same as for news produced completely by human journalists, while 31 percent believe that it is less worth paying for news produced by AI.

In Jornet's opinion, these figures do not necessarily indicate that in Argentina there are more people willing to pay for news produced by AI than in the rest of the countries.

“A news item enriched with AI can undoubtedly be much more valuable, but I also believe, and experience tells us, that the news that people value most are those that have some interesting creative development, an author's signature,” he said. “There are new AI developments that are really impressive, but I still believe that in the work of journalistic and artistic creation, there is still a difference in favor of the human viewpoint.”

For his part, Pino believes that there are other uses of AI that could positively impact media business models, such as the personalization of content adjusted to the interests of each user or the automation of tasks. However, he said that, to this day, the main value of journalism continues to be that added by flesh-and-blood journalists.

“Given the advance of generative AI, journalism must take refuge in its most basic principles, those capable of moving and mobilizing the conscience of citizens, betting on investigative journalism and confronting the powers that be in power to reveal cases of corruption or embezzlement of funds,” Pino said. “AI, in this aspect, is nothing more than a tool that should facilitate the task of committed journalism.”

Translated by Teresa Mioli
Republishing Guidelines