A new report on public attitudes towards AI has found that most people believe human oversight is “essential” for the use of AI technology in the news.
With much of the news industry considering the potential use of AI in content creation, the Oxford-based Reuters Institute has begun researching how audiences feel about AI-enabled journalism.
In a survey commissioned by the institute earlier this year, participants tried out and considered 25 use cases for AI news.
The report, released today, finds that while there is “generalized suspicion and concern” about the automation of journalistic content, attitudes are more nuanced depending on the type of content and the degree of human oversight.
The report states: “Comfort levels with using AI in news vary widely across applications.
“Initial negative reactions to journalists' use of AI were typically based on the assumption that AI would be used for content creation. As participants were exposed to more use cases, their attitudes became more nuanced and, overall, more positive.”
The key factors determining public attitudes are:
● Where is generative AI used in the news creation and distribution process?
● The type of information being conveyed and whether human interpretation or emotion is required or desired.
● The medium itself – text, illustrations, photographs and videos were all viewed differently.
● How much human oversight is required?
The report continues: “Most people believe that human oversight is essential to the responsible application of AI. Participants believed that oversight is a principle of good journalism practice in general, and when it comes to AI in particular.”
“There's been a lot of discussion and hype around generative AI and its use in news, but the perspective of the news consumer is often missing,” said researcher Konrad Calao.
“This study aims to redress that balance by taking a select group of people on a journey to understand for themselves how AI is applied to journalism and form an informed opinion about the potential risks and benefits.
“This study sheds light on when, where and how publishers can safely integrate AI, and conversely, where they need to exercise extreme caution.
“Importantly, it also provides insight into why news consumers show different levels of comfort with different uses in different contexts.”
The full survey results can be downloaded here.
