The blank page remains empty until a writer begins to write. A word processor is a virtual page with a blinking cursor, waiting for the writer to begin. Google is a box with a cursor: Do you feel lucky? ChatGPT is another kind of page, an interface waiting for the writer’s input: Ask it anything. A writer may use a prompt, on paper or on a screen, to begin writing. A sentence emerges from the subconscious, from the interior monologue. We have been taught that a sentence is a complete thought, and the writer builds thought upon thought, sentence upon sentence.

Google, however, answers back. It takes the writer’s words and returns text somehow related to them. ChatGPT requires more: the author must write a prompt instructing the machine, and the machine replies, finding, revising, and composing text guided by that input. Yet this transformation does not bridge the gap between the void of the page and the thought only a human can supply. Whether on a blank sheet or through ChatGPT, the something that comes from nothing always begins in a human mind. While the machine produces sentences, they are not thoughts. They are a pattern that has been probabilistically strung together. It has been assembled from a finite but enormous library of fragments of existing text. This absence of consciousness gives the language of large language models (LLMs) its uncanny quality. The machine’s output may sound human, but it lacks the intentionality born of lived experience and reflection.

Machines have long shaped art. Surrealists toyed with chance through the parlor game of exquisite corpse. Exquisite corpse is a collaborative, chance-based creative game where participants sequentially contribute words or images without seeing the previous parts, resulting in a surprising and often surreal final composition. The OULIPO group crafted self-imposed constraints that forced language into unexpected shapes, treating these rules as engines for “potential literature” beyond individual inspiration. Brian Gysin pioneered a mechanistic cut-up process by physically slicing and rearranging text to disrupt linear meaning and generate unexpected narratives. Collaborating with Gysin, William Burroughs adopted the technique to splice and reassemble fragments, producing the disjointed, hallucinatory structures of The Ticket That Exploded and Nova Express. These movements showed that mechanical and algorithmic methods could augment or even extend creativity and human expression.

The partnership between writer and machine recasts the role of authorship. When working with an LLM, the author acts as an orchestrator or director, guiding algorithms as they generate scenes, lines, and fragments. Like James Michener’s reliance on researchers to supply raw data, or Jerzy Kosinski’s controversial use of uncredited assistants and ghostwriters to shape his novels, the model provides material that the author must refine and direct. It is the writer’s intention, not the machine’s output, that gives the work coherence and voice. Although avant-garde and post-structuralist critics have long proclaimed “the death of the author,” even in the age of LLMs, the author remains very much alive.

Understanding this dynamic is essential. Text generated by an LLM is not plagiarism in the traditional sense, as it produces new sequences of words rather than copying from sources. However, its originality is constrained: the model generates output by statistically recombining patterns from its training data without understanding or intent. As a result, the raw text lacks perspective and meaning. Rewriting is crucial, for it is through the author’s intervention that the material gains coherence, insight, and a distinctive human voice.

The rewriting process demands far more than simple rephrasing. It involves dismantling the machine’s output, interrogating its ideas, and reshaping it into language that truly originates from the writer. This struggle, whether working with raw material from memory, research, or machine-generated text, is the medium through which the author’s voice emerges. Readers often describe the experience of reading as hearing the author’s voice, even when the author has long been gone, as with Jane Austen or Emily Dickinson. Writers themselves frequently acknowledge this process with the saying, “writing is rewriting.” When Truman Capote heard about Jack Kerouac’s method of producing On the Road in a single Benzedrine-fueled session on a continuous scroll, he said, “That’s not writing, that’s typing.” While ideas form the foundation, it is the deliberate crafting of language that transforms thought into an enduring expression of the author’s mind.

LLMs are assistants. They can process vast sets of text, detect patterns, and generate possibilities with a speed no human can match. Conceptually, this is not so distant from freewriting, where the writer dredges the subconscious to uncover patterns and ideas as quickly as the pen can move across the page. Yet writing is more than the rapid production of words. It is the slow, deliberate act of shaping meaning from disorder, of confronting and wrestling with the logic of the text.

Critics such as Q. D. Leavis, in Fiction and the Reading Public, argued that mass literacy and commercial publishing had weakened cultural and literary standards. Today, similar fears suggest that machine-generated text will erase literature, that the writer will be replaced by an algorithmic ghost. Such anxieties underestimate writers’ resilience, a craft that thrives when confronted with new challenges.

Writing is itself a technology, a five-thousand-year-old invention that transformed human thought. The alphabet, the printing press, and the word processor each altered how writers work, yet none supplanted the act of writing itself. LLMs are the latest stage in this long progression. They will change how we write, but because they are incapable of thought, they cannot replace writers.

The writer’s role is clear: to use the machine’s output as raw material and wrestle with it, as they would with a freewrite, until it speaks with their own lucid, human voice. The blank page may respond, but it echoes like a myna bird, repeating without understanding. Turning thought into tangible text still requires a human writer. Writers who embrace these tools as collaborators, not replacements, will shape the future of literature.