By Gabriele Hooffacker and Nicola Moser
Abstract: Generative language models and AI tools have become essential tools in journalism – used in data analysis, research, translation, idea generation, and much more. How will the use of tools like ChatGPT impact the shape of journalism as a profession, and its academic teaching? Analysis of these expert interviews shows that ChatGPT and similar AI tools are already playing a role in academic journalism training. But while university teaching assumes that generative language models will not fundamentally change the shape of journalism, but merely expand it, the practicing expert interviewed sees a fundamental shift in the relationship between editorial offices and audiences. He also describes how the use of AI tools has long become common practice in editorial offices.
Numerous publications have looked at how the use of generative AI platforms, especially language AI, will change the shape of journalism as a profession and thus also journalistic training at universities. Yet the development is so dynamic that studies quickly become obsolete: »The use of AI in journalism is not an isolated or completed topic, but a dynamic and interdisciplinary field that demands continuous and critical reflection, including from the perspective of communication studies« (Wolf 2024: 24). Many possible applications are known, and initial experiences have already been evaluated. »The large language models can be used creatively to generate ideas, for suggestions for article structure, the thread of a story« (Haustein-Teßmer 2024: 75). The ability to use AI in a conscious and controlled way is becoming essential. An open attitude to technology, curiosity, interest, and a use of AI that is both playful and critical must be anchored in journalism training (cf. Hooffacker 2023: 210f.).
»Digitalization has changed journalistic training, careers and processes at regional media companies. But there are deficits,« says Oliver Haustein-Teßmer (Haustein-Teßmer 2024: 68). The young journalists whom Vera Katzenberger interviewed for her dissertation share that view (Katzenberger 2024), valuing training in specific digital tools less than openness to technology. One generally needs to be »very open and flexible towards new technology,« said one interviewee (Katzenberger 2024: 186).
»Looking at the didactics of journalism research drives right into the heart of the debate in journalism research about how it sees itself. This debate has always centered around the relationship between theory and practice. Journalism research has repeatedly been accused (and accused itself) of neglecting its explicit practical orientation,« write Beatrice Dernbach and Wibke Loosen at the start of their 2012-published Didaktik der Journalistik (Dernbach/Loosen 2012: 11).
Is this also true in relation to the use of generative language AI and large language models (LLM) in academic journalism teaching and practical journalism? What do we know about how the profile of journalism as a profession is changing? And what are the consequences of this for academic journalism training?
Method
This paper is based on guided expert interviews that were conducted in 2024 and analyzed in a qualitative, deductive content analysis in line with Mayring. Nicola Moser developed the categories for the interviews’ coding guidelines for his master’s thesis (Moser 2024) and used them as the basis for creating the questionnaire. Eight university lecturers from the fields of journalism, data journalism and media analysis were interviewed. The author then added to and compared the results with an interview with an innovative journalism practitioner, using the free AI software TurboScribe to transcribe the interview and ChatGPT and MS Copilot to help formulate the text.
When selecting the experts, attention was paid to achieving the most balanced gender ratio possible. Three women and five men from communication and media studies agreed to take part in the interview:
- Prof. Dr. Katharina Heimeier, Westfälische Hochschule – Westphalian University of Applied Sciences, Practice and Theory of Quality Journalism program
- Prof. Markus Kaiser, Technische Hochschule Nürnberg Georg Simon Ohm, Practical Journalism program
- Dr. Max Eder, LMU Munich, research associate at the Department of Media and Communication (IfKW)
- Prof. Dr. Andreas Moring, International School ISM in Hamburg, Digital Management program
- Prof. Dr. Markus Behmer, University of Bamberg, Empirical Communicator Research program at the Institute of Communication Science
- Prof. Dr. Volker Markus Banholzer, Technische Hochschule Nürnberg Georg Simon Ohm, focus on Innovation Communication
- Prof. Dr. Magdalena Taube, Macromedia University, Journalism program
- Dr. Theresa Körner, lecturer, research focusing on Automated Journalism
The results of the interviews were collated with the journalist Dirk von Gehlen, who is currently Director Think-Tank at the Süddeutschen Zeitung’s SZ-Institut.
Results of the expert interview
Nicola Moser identified the following eight categories for his master’s thesis:
- K1: Effects on journalism
- K2: Opportunities offered by the language models
- K3: Risks of the language models
- K4: Influence on motivation to learn
- K5: Integration of ChatGPT in teaching
- K6: Changes to teaching content
- K7: Significance of prompt engineering
- K8 Changes in the profile of journalism as an occupation
The interviews were analyzed in line with the categories. The results of the K5 and K7 categories are combined in the description below.
K1: Effects on journalism
The experts agree that ChatGPT has a significant influence on working methods in journalism. The technology is seen as a useful aid that can be used in various phases of production, such as in generating ideas, creating drafts or translating content. However, fact checking remains essential, as the models can sometimes generate incorrect information or even hallucinate. Katharina Heimeier stresses the importance of journalistic responsibility: »In the future, too, one needs to scrutinize the responsibility one holds, the consequences it will have if one publishes this or that.«
Andreas Moring adds that dramatic writing and embedding information in a larger context remain core human skills: »The journalistic human abilities of placing things in context, evaluating them and creating a common thread remain essential.«
K2: Opportunities offered by the language models
There were two topics that the experts interviewed saw as major opportunities: the time saved in editorial work, and multilingualism and the associated internationalization.
Time saving and increased efficiency: ChatGPT can take on routine tasks like composing simple texts, thus creating more freedom for creative and investigative work. Magdalena Taube explains: »It will hopefully create freedom to concentrate on the things AI cannot do.«
Translations and internationalization: Language models make it easier to create multilingual content and expand media companies’ target audience. Magdalena Taube highlights the potential of this for tapping new markets.
In research, the interviewees said, the new tools can do much more. The most common example quoted was the enormous benefit to data-based journalism such as sport journalism. Automation and training artificial intelligence have made exciting results possible for several years now, says Markus Kaiser. A teaching and research project successfully managed to use live ticker results to produce match reports for lower leagues. Markus Kaiser sees this as offering potential to tap niche markets.
K3: Risks of the language models
In analyzing the results, it is noticeable that the experts identify significantly more critical points than advantages. They emphasize that many questions remain unanswered, such as the issue of copyright and source references. These aspects affect not only the legal constraints, but also how trustworthy and traceable the content generated by ChatGPT is.
Specifically, the following risks were identified:
Incorrect information: The danger of hallucinations, in which the model generates false content, means that constant checking by journalists is essential.
Linguistic levelling: Volker Bannholzer sees a risk that the use of AI could lead to simplification of language and convergence on lower linguistic standards.
Fake news and manipulation: Generative language models can be misused to spread incorrect information in a targeted way.
Markus Kaiser asks about transparency: Does content need to be labelled? In her research, Theresa Körner looked at the effect that texts have on an audience that knows how the texts were created. She comes to the conclusion that audiences fear manipulation of opinion and recommends labelling and explanations of how the texts are composed.
Max Eder notes that, in addition to checking the information and sources, AI-generated texts require a great deal of revision and tend to create more work. He also sees »a kind of digital divide« between older and younger journalists. Editorial offices need to find a way to bridge these differences.
K4: Influence on motivation to learn
In her 2023 workshop report, Gabriele Hooffacker comes to the conclusion that »the students test the potential uses with curiosity and enjoyment, and gain from this« (Hooffacker 2023: 211). It is striking that not one of the people interviewed has observed an impact like this among their own students. Three people did not comment on this at all.
Magdalena Taube has observed in her seminars that students rarely use the tools of their own accord. A similar observation comes from Volker Banholzer: ChatGPT is used occasionally but, if at all, for reasons of efficiency.
It is impossible to resolve this contradiction, but it is reasonable to assume that students do not always tell their teachers in detail how they really use generative language AI.
K5: Integration of ChatGPT in teaching and K7: Significance of prompt engineering
According to the interviewees, the integration of ChatGPT into journalistic training is still in its infancy. Initial approaches include teaching the basics of handling the technology, practicing prompting techniques and discussing ethical issues.
Markus Behmer stresses the necessity of integrating the use of ChatGPT into training for young journalists: »It is a journalistic technique that needs to be standard in training.« Handling language AI is a cultural technique, a journalistic technique that needs to be learned, says Andreas Moring: »How do I use it? How do I prompt? Which tool is suitable for what?«
Theresa Körner sees the use of ChatGPT as an opportunity to support students in developing arguments and perspectives: »The models can serve as sparring partners to enhance arguments or develop new points of view.« Prompting plays a key role here, she says.
Katharina Heimeier addresses the topic of digital ethics in her seminars. »Here we discuss whether the Press Code should include this, yes or no? Or we look at specific examples.« She generally emphasizes the media ethics aspect of the topic.
K6: Changes to teaching content
For the university lecturers interviewed, a central consideration is how to adapt examination formats. The most common response is to increase the number of oral examination formats, as papers written at home do not fulfil their purpose if large sections are written by ChatGPT.
Theresa Körner recommends »more examination formats whose aim is for examinees to be able to reflect more effectively and back up their opinion with arguments.« Markus Behmer takes a similar view: At his university, he says, there is no oral defense of final theses. Action needs to be taken.
Will teaching be affected? It is a question that Volker Banholzer asks himself. Experiments are required, for example with students venturing to have texts they have written revised by AI. This should then lead to discussion: Is it OK, or has the text lost its character?
Andreas Moring and Magdalena Taube do not necessarily take a critical view of the use of AI in papers and final theses. However, it is important to be transparent about how the texts came about. Magdalena Taube explains: »At our university, we tell students they are allowed to use AI. But they have to compile a list of AI, like a list of references.« This even includes a list of prompts.
Markus Kaiser notes that the strict separation between print, audio and video as channels is disappearing, as can already be seen on social media channels. Teaching needs to adapt to this, he argues.
K8: Changes in the profile of journalism as an occupation
Here, all the interviewees are in agreement: The profile of journalism as an occupation will not change in the foreseeable future. They expect fully automated editorial offices and multi-channel distribution to play only a minor role initially. As it is impossible to predict how technical possibilities will develop, however, this point of view may change in a few years. Yet all the experts agree that they are seeing a transformation in the everyday working lives of journalists. In a few years, all editorial offices will use generative language models. This is already possible in every part of the working process, they said.
Which skills will young journalists need? Two aspects were named time and again. The first, named by Maximilian Eder and Magdalena Taube, is »digital literacy« – the ability to engage with and scrutinize new technical possibilities. The other is checking sources and questioning information. Magdalena Taube believes that, despite this, the craft of journalism will remain just as important as it is today.
Overall, all the experts interviewed expect generative language models to play an important role in everyday journalism in future. At the same time, critically evaluating and checking information remains a core competency for those working in journalism.
Interview with a practitioner as an extension of the results
In order to reflect the results of the expert interviews onto journalistic practice, questions for a journalistic practitioner and thought leader were developed from the categories. The interview with Dirk von Gehlen initially gives an insight into the working world of journalists, as well as a view of future normality and constant change.
According to von Gehlen, both consumers and publishers use generative language assistants. The first audience tool developed by Süddeutsche Zeitung as a pilot project offered readers a personalized review of the year. It was based on texts from the year 2023 and allowed readers to interact with their content and chat – a first attempt that demonstrated the tool’s potential.
Most of his colleagues, says Dirk von Gehlen, work with DeepL, MS Copilot and other language assistants on a daily basis. Artificial intelligence is used throughout the entire process, from research to presentation to publication, and especially in creating content, structures or initial concepts.
The editorial office is aware of a certain resistance among the audience to texts that are not created by people, he continues. However, this could change as the technology becomes further developed and more widespread, leading to greater acceptance. Transparency is needed.
In contrast, Dirk von Gehlen does not expect prompt engineering to become an occupation. The skills needed for prompt engineering will be essential and go without saying for everyone producing texts, like the use of a search machine.
According to von Gehlen, it is impossible for conventional training to keep pace with the fast development of the technology, with a useful new tool appearing every few weeks. His conclusion: »Probably the most important skill to be taught is handling change and adapting new skills.«
He argues that constant change is part of the occupational profile of journalists: »Every idea of culture, of progress, of education, of scientific research is based on the fact that things change. We have a tendency to always see our status quo as the pinnacle of human achievement and we like it when it stays that way. No, I believe that culture is movement and that journalism, as an accompanying technology or skill of culture, urgently needs to change.«
In many ways, Dirk von Gehlen’s responses match the views of the academic experts. Yet he also sees a drastic change in the job profile and thus new demands on the competencies that need to be taught in journalistic training.
And the innovative practitioner’s considerations go even further: Dirk von Gehlen speaks of a cultural transformation that is being driven by artificial intelligence, as well as by digitalization more generally. He believes that, in the long term, this development will result in a shift in communication between transmitter and recipient.
Another aspect, he continues, is the increasing segmentation of public discourse as a result of personalized content. Media are adapting ever more to the needs of consumers. Generative language systems allow content to be personalized even more.
Von Gehlen uses the analogy of a festival to describe the resulting challenge for journalists: »Today, we see that the public discourse is incredibly segmented, a bit like a music festival where there are thousands of stages. There is no longer a main stage, but lots of side stages on which lots of people are playing who do not know one another at all. You go to Stage C, I’m on Stage A and we know nothing about one another. The groups do not know each other anymore either, because this community connection no longer exists in the same way. In a plural theory of public discourse, this is huge progress, as it allows voices from outside this main channel to be articulated. On the other hand, the strong segmentation of public discourse challenges us. It is being accelerated enormously by AI and thus poses huge cultural challenges for us.«
All in all, Dirk von Gehlen hopes that it will be possible to outsource some annoying tasks to AI in the future. At the same time, there will be new tasks: »It will relieve us of tasks that we used to do ourselves by hand. But it also creates entirely new tasks. If we see journalism as communication, i.e., not just as transporting content, but as true communication with listeners, and respond to one another, we need to ask the question: What defines human communication as opposed to machine communication?« This is what he sees as the core task facing future journalists.
Conclusion
The analysis shows that the use of generative AI tools like ChatGPT is already influencing journalism and has already entered, or will enter, many areas of editorial work. While university lecturers see AI as more likely to expand the profile of journalism as an occupation, rather than change it, the innovative practitioner sees a fundamental shift – especially in the relationship between editorial offices and their audience.
The opportunities the experts see include increases in efficiency, new forms of research, and internationalization through multilingual content. However, they also see grave risks: incorrect information, linguistic levelling and potential manipulation by generated content are major challenges. Fact-checking therefore remains an essential journalistic responsibility.
There were contradictory observations from university teaching. While Hooffacker saw students using AI in a playful and interested way, other interviewees reported more hesitant use. There were also differing views on the adaptation of teaching content, ranging from a stronger emphasis on oral examinations to the introduction of AI evidence in academic work.
The difference between the academic view and that of practitioners was especially clear when it came to the future occupational profile. While university teachers expect to see hardly any fundamental change, Dirk von Gehlen sees a deep-rooted transformation that is being accelerated by AI and digitalization. The advancing personalization of content could lead to greater fragmentation of the public discourse, he says, bringing with it new tasks for journalists.
Comparing findings from academic journalism teaching with the conclusions that can be drawn from how AI is already being used in practice clearly highlight a core challenge: Journalistic training needs to teach not only technical skills, but also the ability to adapt flexibly to the transformation. After all, as von Gehlen stresses, change is an inseparable part of journalism.
About the authors
Gabriele Hooffacker, Prof. Dr. phil., (*1959), is an editor of Journalism Research and teaches in the Media-appropriate Content Preparation department at Leipzig University of Applied Sciences. She publishes the » Journalistische Praxis« series of textbooks at Springer VD, founded by Walther von La Roche (1936-2010), and the series »Leipziger Beiträge zur Computerspielekultur«.
Nicola Moser (M. Eng) studied Book and Media Production (B. Eng.) at Leipzig University of Applied Sciences and completed a master’s in Media Management in 2024.
Translation: Sophie Costella
References
Dernbach, Beatrice; Loosen, Wiebke (2012): Didaktik der Journalistik. Wiesbaden [Springer VS] 2012
Haustein-Teßmer, Oliver: Wie die Digitalisierung die Journalismusausbildung verändert. In: Hooffacker, Gabriele; Kenntemich, Wolfgang; Kulisch, Uwe (eds.): Neue Plattformen – neue Öffentlichkeiten. Wiesbaden [Springer VS] 2024, pp. 67–78.
Hooffacker, Gabriele: How language AI could change journalism training. In: Journalistik/Journalism Research, 6(2), 2023, pp. 189–196. https://journalistik.online/en/essay-en/how-language-ai-could-change-journalism-training/ (28 November 2024)
Katzenberger, Vera: Zwischen Anspruch und Wirklichkeit. Cologne [Herbert von Halem] 2024
Moser, Nicola: Die Auswirkungen von Generativer Sprach-KI wie Chat-GPT auf Berufsbilder im Journalismus und deren Einfluss auf die akademische Journalismuslehre. Unpublished master thesis. Leipzig [HTWK Leipzig] 2024
Wolf, Cornelia: Journalismus und künstliche Intelligenz aus kommunikationswissenschaftlicher Perspektive: Chancen und Herausforderungen. In: Hooffacker, Gabriele; Kenntemich, Wolfgang; Kulisch, Uwe (eds.): Neue Plattformen – neue Öffentlichkeiten. Wiesbaden [Springer VS] 2024, pp. 9–29
About this article
Copyright
This article is distributed under Creative Commons Atrribution 4.0 International (CC BY 4.0). You are free to share and redistribute the material in any medium or format. The licensor cannot revoke these freedoms as long as you follow the license terms. You must however give appropriate credit, provide a link to the license, and indicate if changes were made. You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use. You may not apply legal terms or technological measures that legally restrict others from doing anything the license permits. More Information under https://creativecommons.org/licenses/by/4.0/deed.en.
Citation
Gabriele Hooffacker, Nicola Moser: »I believe that journalism is in urgent need of change«. On the relationship between academic journalism training and journalistic practice. In: Journalism Research, Vol. 8 (2), 2025, pp. 227-236. DOI: 10.1453/2569-152X-22025-15342-en
ISSN
2569-152X
DOI
https://doi.org/10.1453/2569-152X-22025-15342-en
First published online
July 2025