LMU researchers conduct representative survey on the perception and use of AI in journalism.

Professor Neil Thurman and Sina Thäsler-Kordonouri from the Department of Media and Communication (IfKW) at LMU have published comprehensive findings on the perception and professional use of artificial intelligence by journalists.

Although a clear majority of UK journalists (62%) are of the opinion that AI represents a large or very large threat for journalism – and just 15 percent view it as a large or very large opportunity – the study found that most journalists (56%) use AI weekly, while 27 percent use it daily. Only 16 percent said they had never worked with AI. The three most common uses (transcription, translation, and grammar checking) relate to language processing. However, AI is also used for more substantive journalistic tasks: More than a fifth of respondents said they used AI at least once a month for ‘story research,’ while 16 percent used it for ‘idea generation’ and ‘generating parts of text articles.’ The least common use of AI was for the production of videos.

In a large online survey of over 1,000 UK journalists, the Munich research team took a snapshot of the perception and actual use of AI in UK journalism. The results of the study, which was co-authored with Dr. Richard Fletcher, have been published in a report by the Reuters Institute for the Study of Journalism at the University of Oxford.

Who uses AI most often?

The study revealed that the use of AI tends to be higher among journalists who identify as male and younger journalists. The use of AI also rises with increasing management responsibility. However, AI use is not associated with satisfaction: “Frequent users are not more satisfied with the amount of time they spend on complex and creative tasks, such as conducting interviews and doing research,” notes Neil Thurman. “Indeed, those who do not use AI at all are the most satisfied.” Moreover, AI use does not increase the satisfaction of journalists in relation to the amount of time they spend on low-level tasks, as journalists who frequently use AI were more likely to believe they had to work on low-level tasks too often. One explanation for this finding could be how AI use is accompanied by AI-specific low-level tasks, such as cleaning data, creating prompts, and checking AI output, observe the researchers.

AI integration in news organizations

AI integration is more widespread in news organizations that belong to media conglomerates than in independent media. Furthermore, the majority of journalists think integration will increase. Some 40 percent of respondents work in newsrooms that have in-house AI guidelines for most of the queried issues, including human oversight, data privacy and security, and transparency. Around a third (32%) have the opportunity to receive AI training. However, training tends to be more available to journalists working for media conglomerates than for their counterparts in independent media. More than half (57%) indicated that their main news outlet only uses third-party tools.

Overall, the results of the study show increasing use of AI in journalism in spite of reservations and at times strong ethical concerns on the part of respondents. It also reveals a growing gap between independent media and large corporations: “Without substantial change, less well-resourced news media will continue to use AI less often. Moreover, their staff will receive less AI training and will be more strongly dependent on AI tools from third-party providers,” believes Sina Thäsler-Kordonouri.