A few days ago, Nobel laureate in literature Olga Tokarczuk commented that she had used artificial intelligence for the research for her new book. Shortly after, a story that won the Commonwealth Prize received numerous accusations of having been produced with AI. Now the issue is closer to home: after being accused of plagiarism on X, Antoni Gelonch has had to admit that he has used AI to write an article.The Turing test has never been on everyone's lips so much: was this written by a human mind or a digital mind? It is not surprising, however, that it is becoming increasingly difficult for us to distinguish between one type of mind and another. After all, digital minds are built from everything that humans have created over the millennia. In recent decades, we have placed an excessive emphasis on productivity and have increasingly prioritized systematization and structure, even in spheres considered most creative. There are countless guides and manuals in English on how to write articles, essays, poetry, narratives, novels, or doctoral theses, on how to conduct research or give talks that captivate the audience. We have measured and limited time and characters, we have designed the most suitable strategies to execute each task, we have created courses for absolutely everything. AI highlights that much of what we do is –or has become– remarkably mechanical: that's why machines can pass themselves off, with some ease, as us. As I was pondering all this in recent days, an article written by Víctor Català in 1927 came to mind. In the text in question, Català comments that criticism had considered that Drames rurals (1903) had shattered “a multitude of conventions, which, accepted through inertia and respected out of inertia, become a canon, a formula obeyed by the general public, who end up finding in them a kind of mattress on which to rest and mold their own taste”. Well, now it seems that what they praised about Víctor Català – the ability to identify and avoid inertia – will surely be one of the skills that humans will have to cling to the most.
When in 1950 Alan Turing writes the famous article in which he theorizes about computational machinery and intelligence, he takes poetry as an instrument for measuring humanity. Can a machine write verses that move us? With this test, Turing made it clear that poetry is not a purely strategic game, like chess. At the same time, he established a clear analogy between the learning processes of machines and the act of literary imitation based on patterns.And here is another key point to understanding our current moment: if we find it difficult to distinguish between text produced by a machine and text produced by a human, it is because both automatisms and imitation are part of our way of existing. Colin Burrow reflects on these issues in Imitating Authors: Plato to Futurity, a study in which he argues that it is a challenge to mechanically replicate a type of imitation like ours. Because, on the one hand, it is extremely complex, given that it has many different layers, and because, on the other hand, it is intrinsically anarchic. Humans learn a series of rules from other humans, but we also learn ways to transgress them and, through inference, we create new ones. We could say that, in a way, we have an unavoidable perverse, subversive instinct. AI, on the other hand, has rigid and – I will say the obvious now – mechanical behaviour. One piece of evidence that, for the moment, AI does not share the subversive spirit of humans is that it does not understand our sense of humour (humour is based precisely on the subversion of patterns). (humor is based, precisely, on the subversion of patterns).godfathers of AI – explained it very well in a tweet. AI compensates for its shortcomings (lack of common sense, understanding of reality, etc.) with an enormous amount of “declarative knowledge”. It has so much material to choose from and sift through that it doesn't need a creative capacity like Víctor Català's. In a moment, it can regurgitate a jewel that had been buried among so much text. When, in August 2025, OpenAI launched ChatGPT 5, it assured that its “integrated thinking system” put “expert-level intelligence within everyone’s reach”. They did not add, of course, the chaos that language models would entail. From universities to primary schools, from articles to scholarship applications or literary awards: we have to rethink everything. For the moment, the promise that AI would take our jobs makes us laugh. Because it has backfired on us.
