Artificial intelligence has become an inseparable part of modern life. The defining technology of our era is now routinely compared to the invention of the computer – or even to the advent of the printing press more than 500 years ago. Even to non-experts – and I count myself among the vast majority of people in that category – it is increasingly clear that these new digital technologies are reshaping virtually every sphere of human activity. Their impact is being felt across more or less all aspects of the lives we lead.
The changes are particularly dramatic for those engaged, in one way or another, in the production of content. That content may take the form of images, videos or films. The same is true of music, but above all of texts of every kind. In the multimedia world we inhabit, the written word remains – perhaps I am old-fashioned in this regard – the highest and purest form of journalistic communication. At the very least, it remains the medium best suited to conveying complex arguments and sophisticated analysis.
And that brings me to the reason for the reflections that follow. As a commentator and political analyst, driven by the admittedly ambitious aspiration to reach for my keyboard only when I have an original message to share, the question is no longer whether I use artificial intelligence in my work. The question that increasingly occupies my mind – and I suspect that of many others – is how I use these digital tools. And here, I should add, things become more complicated: Where are the limits of their use? I am not referring to the technical limitations of these new technologies. For if the experts are to be trusted, the possibilities appear virtually limitless.
The real issue, rather, concerns questions that touch on the ethics of writing and, by extension, the very identity of journalism at a time when the profession is undergoing a radical transformation. This debate extends well beyond the personal concerns of media professionals. Ultimately, it is also about the quality of the media in general – and therefore about political culture itself, with far-reaching consequences for the democratic order in which we aspire to live.
In my native Germany, the debate I am addressing here reached a broader audience after a leading daily newspaper abruptly barred one of its most acclaimed commentators from writing for its pages. The trigger for the “scandal”: The revelation that – and here I quote from Der Tagesspiegel’s statement – the former publisher and editor in chief had “opinion pieces written by artificial intelligence.” In its explanation to readers, the newspaper acknowledged that artificial intelligence is also used in its own newsroom as a “tool” that “helps us simplify and improve individual editorial tasks.”
However, Artificial Intelligence, the newspaper’s management went on to say, is not a tool that should be allowed to “take over the core of our work.” Then comes a sentence that, in my view, is of particular importance – and one that is central to my own understanding of journalism: “Journalistic judgment, the weighing of information, analytical interpretation and the crafting of language must always remain the responsibility of the author.”
The former publisher, caught violating the rules he was expected to uphold, promptly acknowledged his mistake and publicly apologized. In doing so, he conceded that the very least he should have done was inform readers that the text published under his name had in fact been generated by software.
The “Casdorff case” – so named after the publicist who fell from grace – is of broader significance because it ultimately concerns more or less everyone who produces content for publication. The crucial question is not: Did I use artificial intelligence? The real question is: How did I use it? And – here lies the heart of the matter – to what extent is the final product the result of my own intellectual effort and independent judgment?
Casdorff failed because he ultimately placed his name on a text generated by an artificial intelligence program. Readers were therefore entitled to assume that what they were reading genuinely originated in his own thinking. But that was not the case. The creative work did not come from Mr. Casdorff, who was identified as the author, but from a lifeless machine.
In this instance, the newspaper’s editors were convinced that a red line had clearly been crossed. What we are dealing with is a modern form of plagiarism made possible by artificial intelligence. Yet the question inevitably arises: What remains permissible in this new world of seemingly limitless technological possibilities? And is there, in fact, any universally valid answer to that question?
To provide clarity and guidance on these questions, media organizations have published extensive guidelines governing the use of artificial intelligence. In Germany, eight out of ten newspaper publishers now have such guidelines in place. Some even require their staff to use these modern tools designed to streamline and rationalize editorial workflows.
AI-powered applications have become an indispensable part of everyday journalistic practice. Tasks that, not long ago, were handled by search engines, spell-checkers and translation software are now performed – faster, more effectively and within a single platform – by AI assistants. Particularly powerful, and therefore especially popular, are applications that can generate headlines and text summaries at the click of a button. In most cases, the results are ready for publication.
At this point, the negative side effects of technological progress begin to come into view. Summarizing texts and crafting strong headlines rank among the most intellectually demanding tasks in any newsroom. If these responsibilities are routinely and almost automatically delegated to artificial intelligence, the result is intellectual complacency and a gradual erosion of critical skills. A creeping dependence on the machine takes hold – one that can function much like a drug. The danger is that core journalistic competencies will erode over time or even disappear altogether.
And so I arrive at a concluding observation – one that was most certainly not generated by artificial intelligence. There is no alternative to engaging proactively with AI and integrating it into everyday professional workflows. What matters is that the author of a text regards the software as a tool rather than as an automated – and therefore dehumanized – generator that does all the work. Editorial responsibility rests with human beings, not with an anonymous system.
And finally: Should there ever be a lack of time – or of the intellectual capacity required to do the work independently – the least the public is entitled to expect from an author is disclosure that the text was generated by a machine. In that spirit, I should tell readers that, in translating my thoughts from German into English, I occasionally turned to a digital translation program for assistance.
Dr. Ronald Meinardus is a political analyst, commentator and Senior Research Fellow at the Hellenic Foundation for European and Foreign Policy (ELIAMEP).
