Subscribe
How should researchers, journalists, and authors approach AI?
How should we view authors who put their name to a text they did not write themselves—or one where AI, at the very least, authored certain parts?
Sweden has now seen its first case involving an established researcher who, according to investigative journalists, is alleged to have used AI in their writing.
The case concerns Emma Engdahl, a professor of sociology at the University of Gothenburg. According to journalist Anna Axfors, writing for the magazine “Kvartal”, Engdahl has published several texts—and even books—that were actually written by AI.
Axfors grew suspicious while leafing through a recently published book by the sociology professor, in which the author repeatedly employed the same rhetorical patterns in her arguments. It is well known that AI tools like ChatGPT and Claude tend to produce easily digestible prose, often built upon simple—and sometimes illusory—contradictions; for instance, asserting that “A” is not merely “A,” but also “B.”
Professor Engdahl’s new book was a collection of essays titled “Lektioner i kärlek” (“Lessons in Love”). However, according to journalist Axfors, the book should have been titled “Lessons in ChatGPT”. In her article for Kvartal, Axfors cites passages from Engdahl’s book, demonstrating how it is fundamentally built upon banal contradictions—presented through a fragmented style of language intended to convey a sense of literary quality.
One of the passages from the book that Axfors cites reads as follows:
“They say we fall for looks—but stay for something more.
And that is true. What draws us in is often not what is visible, but rather what is barely noticeable.
Something felt, rather than shown.
A lingering glance.
A gesture imbued with reflection.
Shoulders relaxing in a sense of security upon being truly seen.
A scent that makes memory stir—evoking childhood, intimacy, or something long missed.
We are drawn to the subtle—not that which shouts, but that which whispers: Here is someone who has lived.”
As Axfors writes: “Engdahl employs, almost mechanically, the same rhetorical device over and over again—asserting what things are not, only to offer an equally vacuous answer as to what they are.”
It may be of some comfort to those concerned about the impact of AI on research and literature that Professor Engdahl’s use of artificial intelligence does not appear to have lent the work in question any intellectual or aesthetic merit. The text comes across as banal in its content and repetitive in its style.
According to Axfors, the analysis establishes that 88 percent of the text in “Lessons in Love” was authored by AI. In the works the professor published prior to the advent of AI, there is no trace of such technology whatsoever. This is, of course, logical. The problem for Engdahl is that this serves as clear evidence that the 2025 book is indeed AI-generated, and that the analysis is accurate. Because if anyone were to claim that the analysis could be mistaken—or that Engdahl’s own writing style happens to resemble what a machine might produce—it remains curious that no such results appear when examining her earlier texts.
But what does Emilia Engdahl herself say regarding these accusations? When Anna Axfors confronts her with the findings of the analysis, Engdahl replies in writing: “Every thought and every word in my books and articles is my own, and I take full responsibility for them. I can justify and elaborate on my theories and lines of reasoning because they are built upon the knowledge I have cultivated. I also engage in these discussions regularly within academic settings—at conferences, in seminars, and through my teaching. A text does not exist in an isolated universe; it is discussed, cited, and continues to live on—surely, at times, even within various AI tools.”
She further adds: “To this, I would like to add that we should move past the debate over whether or not a text is AI-generated. Evaluate the content of the text. Period. The purpose of AI is to enhance human capabilities—across all sectors of society. It is, therefore, a rather curious notion to suggest that we should avoid using AI. Doing so would be counterproductive and, for those of us in the social sciences, would pose a significant risk.”
In short, Engdahl attempts to trivialize what has occurred. It is her own reasoning that is presented in her books and texts; it is the content that matters. It is time we stopped viewing AI as a problem and instead learn how to utilize this new technology.
Yet, no one has been swayed by these arguments.
AI cannot be compared to seminars or conferences. It is a machine-driven agent capable of generating texts to which researchers and authors can subsequently attach their own names. Having AI compose a text on a subject one may have developed oneself—perhaps even employing a unique terminology of one’s own devising—is not the same as writing the text personally. A researcher publishing a piece is expected not merely to present research or findings, but also to engage in reasoning, employing persuasive arguments and precise, accurate language.
Another text by Engdahl that Axfors examined was an essay published in August 2026 in the culture section of the esteemed daily newspaper “Svenska Dagbladet”.
Upon the revelation regarding the use of AI, Svenska Dagbladet’s culture editor, Ida Ölmedal, decided to unpublish the article in question. In a statement in the newspaper, Ölmedal explains that Svenska Dagbladet conducted its own review of the text, followed by a fresh editorial assessment. The conclusion reached was that the text should be unpublished.
Ölmedal emphasizes that essays in Svenska Dagbladet’s culture section are expected to be written by human beings. “In these daily essays, researchers, authors, and other selected contributors delve into subjects in which they are personally and deeply grounded. It is essential that the person who articulates ideas, chooses words, and defines concepts is the author themselves. Language and thought are not so easily separated.”
In short, articles in Svenska Dagbladet must be written by human beings who engage in their own reasoning and communication. Vibrant and intellectually responsible individuals should write the articles to which they attach their own names.
It may seem strange that an experienced professor would expose herself to the risk of public humiliation in the way Emilia Engdahl has. Engdahl is employed at a major university—an environment where there is surely an active discussion regarding students using AI to complete their assignments rather than doing the work themselves. Furthermore, one would assume that faculty at the University of Gothenburg employ tools to detect AI-assisted cheating. Yet, Engdahl herself has fallen into this very trap.
It is also curious that an established researcher—an author of several books—would see any benefit in letting a machine write her work. Does the writing process not offer Engdahl any intellectual pleasure? Creative individuals generally need to exercise their creativity to thrive. In this case, Engdahl allowed a machine to write text that she claimed to have authored herself. She put her name to a book and several articles that she did not actually write. Bizarre!
The question of how AI will be used is complex. How will we perceive research and literature when machines can accomplish things previously achievable only by humans? For now, however, AI-generated text appears to be characterized by a certain banality. Moreover, it seems that we humans remain committed to an ethic that holds that it is the human subject—the human being—who should create and write both research and literature.
Naturally, AI can serve as a technical aid in research. Many already use it for tasks such as searching for references or exploring straightforward research questions. But allowing AI to draft the presentation of research or letting AI write a text with literary ambitions—that is something we are not yet prepared to accept.
The machines have not yet triumphed over humans.