Early AI literature is caught between two forces: the disdain of traditional gatekeepers, who dismiss it as soul-less, and the self-imposed concealment of its authors, who often hide the technology's role by mimicking human predecessors instead of forging a new literary future.
We are living through a moment of thrilling, chaotic experimentation, akin to the dawn of a new art form. Just as inventors and visionaries in the late 19th century fumbled with clunky cameras to see what “moving pictures” could possibly be, we are now probing the capabilities of Artificial Intelligence, specifically Large Language Models (LLMs), to generate literature. This is not a period of polished masterpieces or established canons. It is a fairground of ideas, where the primary goal is to demonstrate, with wonder and enthusiasm, that the machine can do creative writing at all.
The new narrative authors that use LLMs have a hard time establishing their own cultural legitimacy, often being looked down upon by the successful authors, publishers and hobbyists of traditional writing.
The work is frequently dismissed as gimmicky, soulless, or technically flawed—a “literature of attractions” rather than a legitimate new form. There is a parallel desire among some traditional gatekeepers to either denounce AI authors outright or, when they themselves experiment with the technology, to downplay its role to avoid association with a stigmatized, nascent medium. Just as early narrative cinema presented itself as a mature art form distinct from the fairground origins it both rejected and relied upon, today’s literary establishment often frames AI-assisted writing as a threat to “real” literature, even as the experimental, attraction-driven phase of AI creativity unfolds—a phase that history may one day recognize not as a primitive error, but as its own essential and revolutionary chapter.
Early AI literature is fundamentally an exhibitionist art. Its foremost drive is to show that it can show, current AI experiments are displays of possibility. An LLM might be prompted to write a poem—or a Haiku even—in the style of a haunted vacuum cleaner, or a short story where every third word is a color. The “story” is a mere pretext, a string upon which to hang a demonstration of stylistic trickery, conceptual novelty, or sheer scale.
Yet, a significant tension defines this early period. While the potential to explore entirely new literary territories is vast, many AI authors, seeking acceptance or fearing the “gimmick” label, work hard to conceal their use of the technology. They do this by training models to closely imitate established styles and replicate familiar narrative patterns from the human past. This drive toward seamless imitation is a retreat from the medium’s revolutionary call. Instead of masking the machine’s voice to sound like a human predecessor, the most compelling work uses the “attraction” to point the way into a unique future, developing forms and voices that could not exist without the AI.
Of course, as the technology becomes more normalized, a narrativization of AI literature is already underway. We see this in the rise of sophisticated, AI-assisted (web) novels that follow strict genre conventions, or in corporate experiments using LLMs to generate endless variations of children’s stories. These forms bind the signifiers of the medium to the narration of stories and the creation of a smooth, self-enclosed world. The look at the tech stack becomes taboo; the obvious seams and tricks are sanded down in favor of psychological realism or plot efficiency. This is the inevitable rise of the “classical”
The radical potential of this early phase lies in its carnival spirit, its willingness to prioritize impact over illusion, address over absorption. To view today’s experiments solely through the lens of what they frequently lack—narrative sophistication, emotional depth, canonical status—is to miss their historical moment and unique power. We are not witnessing the awkward birth pangs of a new realism, but the vibrant, one-time-only spectacle of a literature of attractions. Its energy is unexhausted, pointing toward possibilities that a premature focus on storytelling alone could never reveal. The roller coaster is the point, not the destination it might (or might not) lead to.