AI in journalism: opportunities and limitations of use in media organisations

Munich, 23 January 2023

Artificial Intelligence (AI) can support balanced, trustworthy journalism in the digital age. Self-learning AI systems reduce journalists’ workload when it comes to research, circulation and producing media content, thus creating scope for high-quality journalism. At the same time, the use of AI poses challenges for media professionals in their everyday work life and for the open-minded formation of opinions in a democracy. In a recent white paper, the experts of Plattform Lernende Systeme, highlighted how AI tools can be used responsibly in journalism and outlined the conditions to be created to facilitate such use.

AI systems can independently create reports, images or videos, personalise news consumption and assist with research by analysing large volumes of data. According to the authors of the white paper “Künstliche Intelligenz im Journalismus. Potenziale und Herausforderungen für Medienschaffende”, (“Artificial Intelligence in Journalism. Potential and Challenges for Media Professionals”) editorial teams are already using the technology to automate the writing of reports on the weather, sport or traffic, to sub-title audio or film recordings and to tailor articles to users according to their geographic interests. AI can relieve journalists of tedious but necessary tasks, enabling them to devote more time to creative journalistic work, such as conducting interviews or doing investigative research. However, according to the white paper, it is important not to overtax media professionals, as could happen if they are overwhelmed with demanding assignments or in the event of staff cuts.

The use of AI will radically change the world of media, and we’re not talking about the distant future here, as is clear from the recent debate surrounding Chat GPT. Editorial teams should use AI as a tool for the benefit of journalists as well as media audiences. At the same time, media professionals, media watchdogs and politicians must steer and frame AI-based journalism so that it can continue to fulfil its democratic role.

Christoph Neuberger, professor of journalism at the Freie Universität Berlin and Director of the Weizenbaum-Institut as well as a member of the “Law and Ethics” working group of Plattform Lernende Systeme

Editorial oversight remains essential

AI systems can support journalistic balance in news coverage by detecting editorial bias in the selection or prioritisation of items. For instance, they can spot whether a male expert is asked to give a technical opinion more often than a female expert. On the other hand, AI-generated articles can also reproduce hate speech and society’s stereotypes that can be found in the texts on which the AI systems were trained. Another of the challenges presented by the use of AI in media is the workload and expense involved in developing an in-house AI solution. If editorial teams use external solutions, however, they are generally making themselves dependent on tech companies, the majority of which are based in the US or China and may have lower standards of plurality or fairness. According to the authors, as a backstop in AI-based journalism, editorial oversight is essential to ensure that the quality of the AI-assisted articles is up to the standards of journalism.

In order to exploit the opportunities that artificial intelligence affords for a kind of journalism that is balanced and promotes democracy, the experts of Plattform Lernende Systeme recommend the responsible use of AI-based media content and AI tools both by editorial teams and by media audiences. They advise that all media content produced with the aid of AI be identified clearly. They also recommend that AI systems for journalism be certified and that journalists as well as workers in the watchdogs be qualified in the use of AI. In addition, the authors call for press law and copyright law to be brought into line with AI-based journalism and for access to data for editorial staff to be improved; for instance, by having open data portals in state agencies.