Wie verändert sich unser Vertrauen in den Journalismus, wenn KIs mitschreiben? Kevin Hermeneit hat sich in seiner Masterarbeit damit beschäftigt. Im Interview spricht er über die Ergebnisse seiner Studie zur Akzeptanz von KI im Journalismus, über Chancen im redaktionellen Alltag und darüber, was Transparenz bedeutet.
ARIC: You conducted studies on AI and its acceptance in journalism as part of your master’s thesis. Why do you find the topic so interesting?
Kevin Hermeneit: I strive to ensure that everyone in Germany can experience and consume independent journalism. Whether that’s as a writer or as someone who ensures that it works on a commercial basis. For me, it’s more the latter. And I am aware of the extent to which AI is already being used in our newsroom.
One of the biggest fears in society with regard to AI is the distortion of content and thus an indirect and sometimes direct influence on democracy. And that’s where I asked myself the question: It doesn’t quite fit together – how do you identify potential strategic errors? How do decision-makers and editors-in-chief actually think about AI in journalism? And how does this fit in with how readers currently view AI, i.e. how society views it? And are there any conclusions to be drawn from this?
What is one of these contrasts, this gap in perception that you describe?
There is a clear fear in society about AI: the distortion of content, the influencing of images, videos and fake news. And at the same time, the media are the ones who are actually supposed to give us the right information – according to certain ethical principles, according to journalistic diligence, so that I get information that is always truthful.
If AI is used now, does that fit together? I know from working in our company that it can go together. But you have to take the general fears seriously.
Which groups of people have more or fewer fears?
In principle, it was to be expected that the older target group would be more skeptical about the use of AI than younger target groups. Interestingly, however, the older target group is influenced by their immediate social environment. If I were to speak to my grandfather now and he says: “AI, it’s all humbug, it’s all magic.” And as a grandson, I say: ” Grandpa, watch out, it’s not that bad. Let me show you how it works” then it will be accepted more quickly than if some institution or company were to say so.
The classic newspaper reader, even if it is an e-paper, is not 20-30 years old. On average, they are 40-50 years old, but depending on the newspaper title, this can be higher. And then the question arises: Can companies or editorial teams make a certain effort to promote acceptance through transparency? The answer would be based on the data: No, it wouldn’t really be that, there are completely different levers and they tend to make use of the social environment.
Another point is that, according to the information from my data sets, the younger target groups have a higher level of acceptance, but also skepticism. They are more likely to be influenced by social assessments, the basic social acceptance of AI. That is a completely different lever again.
The interview partner
Kevin Hermeneit is a trained media businessman and management consultant at NOZ/mh:n MEDIEN and deals with strategic issues at the interface of digitization, media and business models. He studied Master Business Consulting & Digital Management at the FOM University of Applied Sciences.
What does transparency mean in practice? How can media professionals make companies (more) transparent?
There are different perceptions among the experts in the industry that I interviewed. Some of them refer to the basic promise of transparency: You will receive truthful information from us, regardless of whether an AI was involved or not. If we no longer keep this basic promise, then we have lost. It is completely irrelevant whether I then label an AI with an abbreviation or not, or whether I label individual articles. There was also the thought: if I make some things transparent now, won’t that wake sleeping dogs? Won’t I create this fear and mistrust, even though we editors know that it’s not actually that dramatic? And then this transparency would have a negative effect.
When I carried out my quantitative study, I asked about different levels of transparency, with individual articles also graded: Was produced by machine or Was produced by machine under the responsibility of Max Mustermann and also the third level Was produced by machine under the responsibility of Max Mustermann and checked for accuracy. The greatest trust is of course where the accuracy is checked. There is a considerable difference between machine-generated and machine-generated under the responsibility of Max Mustermann. The human component is the key. And you only get different results as to whether this can be monetized. There is a basic willingness to pay, but this is not the same as for a man-made article.
Where is AI used in editorial offices?
Transcribing, for example, like this interview. If we use an AI to do this, you save a lot of time, which you can invest in high-quality journalistic work. Editing, shortening and correcting texts for tailoring to a print product or even SEO-optimized headlines. There are AI tools that have been intrinsically created and adapted by employees, you enter a text and produce an SEO-optimized headline. A colleague from the editorial team has created a tool that can be used to create smaller reports such as police reports or similar via an AI. You enter a prompt and then the article, including the headline, is created in the right size and quality. As an editor, you know exactly what you have to write so that the AI can work. At the moment, it’s primarily about optimizing work processes in the editorial department by reducing repetitive work steps. However, there are more and more depths of application elsewhere, for example entire AI authors. The author is an AI is also declared in the same way and you know exactly when a text comes in that it is an AI.
What is the mood within the journalistic profession?
Essentially, everyone is rather confident. Many editors-in-chief and AI managers have given me the feedback that they are keen to drive the topic forward, but that they also say: we need to get this speed up as quickly as possible. There are many voices saying, We can only really achieve this together, because in the past, we have always been far too slow in this industry. Maybe we can pick up the pace if we join forces and share our learnings.
Basically, editors always have a purpose to write and put in the best possible work, and you can do that by gaining time like this
In addition to writing, we have been talking about personalization in online applications for years. Of course, this is also a huge field in journalism.
And we also need to talk about research. Researching certain things in certain databases is a huge effort. Sometimes you have to meticulously pick out the information you need. If you go into investigative journalism, an AI can help tremendously with searching through archives. This is also something that can improve the product in the end.
I am surprised by the positive attitude towards it. I would have thought the mood was more conservative.
We have also learned a lot from the past. If you look at how late some media companies got on board with social media and how late some have been to other developments in the course of digitalization since the early 2000s, you realize: either we develop now or we miss out.
Accordingly, this is a spirit of optimism, with the necessary respect. The idea that there is no alternative to independent journalism is clear. It should be continued and, accordingly, we want to do everything we can to take this with us and use it to our advantage.
No negative voices at all?
Of course, any technologization is always accompanied by the issue of job losses. This is a point that I have also taken up as part of my work. I’ve heard the phrase several times, which I think is also true: AI won’t replace your job, but AI will replace the job of people who can’t handle AI
How can I get everyone on board? How can I take away everyone’s fear? This is of course a process that is not easy.
A colleague of mine has just been to a summit. They were discussing a dystopia: What will actually happen if AI completely changes the internet interface as we know it today? We also know, for example, that AI can suddenly become dangerous for Google. AI platforms crawl information from websites and news pages. So why should I still go to a news site? Why should I still have a subscription on a news site? You can monetize reach, convert it into advertising revenue or advertising analyses. If they disappear, your entire business model may no longer be economically viable.
In the end, perhaps the winner will be the one who still manages to do things so well that people go to the website?
Perhaps this also means that a defiant movement is emerging. There are people who know that journalism is a cornerstone of a functioning democracy. That’s rather unimaginable from my perspective, but you could take it a step further and perhaps create a renaissance effect or something similar? Thinking very far ahead.
There is often a counter-trend to major trends.
Humans are a decisive factor in the whole process. Also from an editorial point of view: If I make mistakes that an AI makes because I didn’t even look at it, and I now publish something in a political context, this can of course have a considerable social impact.
Everything I say here reflects large parts of an industry that I personally find very visionary. These are all supervisory people. They have a great drive to say: I know about the responsibility of journalism. You can feel that in all the interviews you conduct. Which is why I personally believe that the industry has a lot of opportunities to adapt to the speed of change and to manage to transfer AI cleanly into journalism. That AI doesn’t immediately mean: this is the downfall of journalism.
Interview: Sabrina Pohlmann
With our interviews, we want to introduce you to different perspectives and players in the field of AI. The positions of our interview partners do not necessarily reflect the positions of the ARIC.
Further interviews:
