Research in Artificial Intelligence has entered a phase of acceleration that we are all witnessing with astonishment—including the very researchers who work in the field. In the areas of computer vision and language, one of the most surprising aspects is how easily artificial brains tackle tasks in which creativity is an essential ingredient. In the field of creative writing, Large Language Models (LLMs) have matched and even surpassed human creative ability in some tasks. But are they truly creative when writing, or do they merely repeat the clichés on which they have been trained? Do they have a style of their own? Can they really be compared with the best human writers? In the talk, we will discuss LLMs' creative potential and intrinsic limitations, especially in the field of creative writing. We will place particular emphasis on a series of experiments carried out at UNED, including a duel between GPT-4 and one of today’s best novelists writing in Spanish, Patricio Pron. This experiment is inspired by other contests between Artificial Intelligence and human champions, such as Deep Blue versus Kasparov and AlphaGo versus Lee Sedol.