This article was written by a human. Will this soon be the exception?
When people talk about how artificial intelligence (AI) is changing journalism, they almost always mean AI systems like ChatGPT – so-called large language models that offer journalists completely new possibilities.
The London School of Economics recently conducted a global survey of over 100 large and small editorial offices. More than three quarters of those surveyed said that they already use artificial intelligence in their everyday editorial work.
“We are still at the beginning”
Swiss media companies are also relying on the technology – albeit cautiously, says Thomas Benkö, who examines the possible uses of artificial intelligence at Ringier: “We use AI, for example, to subtitulate videos or to transcribe long interviews.”
Areas in which artificial intelligence is only used to provide support and does not create content itself. “We don’t come to the office in the morning, press a button and then everything is done – we’re still at the beginning,” laughs Benkö.
AI helps with a lack of resources
Artificial intelligence is also used to support the NZZ media group, CH Media, TX Group (formerly: Tamedia) and SRG. For example, to make suggestions for titles and leads, to send personalized newsletters and push notifications, or to write radio broadcasts.
New guidelines for dealing with artificial intelligence
Open box Close box
Ringier, the “NZZ”, CH Media, the TX Group and the SRG have already committed themselves to the responsible use of artificial intelligence in their journalistic guidelines or are in the process of adapting these guidelines accordingly.
And just now the publishers’ association has Recommendations for action published on the use of AI systems. This also includes the requirement that it must always be declared if content was created solely by an artificial intelligence.
Fed with appropriate material, the AI can also write certain texts itself, says Thomas Benkö: “We are trying to have the AI write match reports based on football data. This means we can also cover games from lower leagues for which we otherwise lack the resources.”
It is important to Benkö to emphasize that none of these texts are published without someone checking them.
AI texts riddled with errors
Things look different in the USA: there have different media In recent months, texts have been published that were only created by an artificial intelligence – sometimes riddled with errors and without any indication that no human wrote them.
Content from established Swiss media that is published solely by AI without human control does not exist today and will not exist in the future.
Will this also happen in Switzerland in the future? Stefan Wabel, managing director of the Swiss Media Publishers Association, says no: “The established Swiss media will not publish articles that are automatically generated and published without human control.” There is no such thing today and there will be no such thing in the future, Wabel assures us.
But he points out that smaller platforms could very well spread such content one day. An assessment that Susan Boos also shares. The President of the Swiss Press Council fears that smaller online media could be less careful in this regard: “You will have to watch closely to ensure that they do not publish things that are actually not possible.”
The audience doesn’t want AI texts
It could be worthwhile for media companies to continue to use artificial intelligence cautiously in the future. According to a current, representative survey by the Public and Society Research Center at the University of Zurich, acceptance of AI-generated content among the population is low.
Less than a sixth of respondents want to read articles written entirely by artificial intelligence. And only nine percent of them would be willing to pay for such content.