Artificial intelligence can help writers overcome creative barriers, improve their prose and explore new ideas. But research suggests that using it effectively requires something no machine can supply on its own: human judgment.
AI has transformed the writing process, allowing authors to brainstorm ideas, develop outlines, revise manuscripts and experiment with different narrative approaches. Yet the growing use of these tools has also raised an uncomfortable question: Does AI make writing easier, or does it simply change the nature of the intellectual work involved?
A recent study by researchers at Iowa State University challenges the popular notion that AI is merely a shortcut for students unwilling to do their own work. Instead, it suggests that effective AI-assisted writing requires experimentation, expertise and personal agency. The technology may accelerate the production of words, but the responsibility for deciding what those words mean remains with the writer.
The distinction matters not only in classrooms but also in journalism, literature and professional communication, where originality, accuracy and individual expression remain essential.
Beyond the Myth of the Lazy Shortcut
The study, conducted by Abram D. Anders and Emily Dux Speltz, identifies three important concepts that shape how students learn to write with artificial intelligence: experimentation, expertise and agency. Rather than simply accepting machine-generated responses, students must learn to evaluate suggestions, recognize weaknesses, refine their instructions and make deliberate editorial decisions.
Writing with AI therefore introduces a different kind of intellectual challenge. The writer becomes not merely a producer of sentences but also a critical reader, researcher and editor of material generated by a system that can sound authoritative even when it is mistaken. A convincing paragraph is not necessarily an accurate one. An elegant argument may contain unsupported assumptions. And a beautifully constructed sentence can still fail to communicate an original idea. The essential skill is knowing the difference.
The Writer Who Refuses to Be Called a Cheat
French writer Sébastien Bailly has challenged the literary world’s tendency to divide authors who use artificial intelligence into two categories: those who remain innocent of the technology and those who supposedly cheat by employing it. Writing in Le Monde, Bailly argues that this division leaves little room for authors who openly acknowledge using AI while retaining creative control over their work.
His argument points to a fundamental misunderstanding of literary creation. Generating a manuscript is not the same as writing a book. A machine can produce hundreds of pages of grammatically competent prose, but a novel requires far more than grammatical competence. It needs observation, imagination, conflict, memory, narrative structure and a distinctive understanding of human behavior. The writer must decide which characters deserve attention, which scenes belong in the story, where the narrative should accelerate and when silence communicates more than dialogue.
AI may offer alternatives, but it cannot relieve the author of the responsibility to choose among them. Indeed, one of the most demanding aspects of working with artificial intelligence may be learning when to reject its suggestions.
Leveling the Creative Playing Field
One of AI’s most promising contributions is its ability to help people who struggle with particular aspects of writing. Some writers possess compelling ideas but have difficulty organizing them. Others face linguistic barriers, problems with structure or the intimidating experience of confronting a blank page.
AI can help transform scattered notes into outlines, suggest ways to clarify an argument or identify passages that need further development. For writers working in a second language, it can also help improve readability without necessarily changing the underlying ideas.
Research published in Science Advances in 2024 offers evidence of this potential. In an experiment examining AI-assisted short-story writing, researchers Anil R. Doshi and Oliver P. Hauser found that access to AI-generated ideas improved evaluations of creativity, writing quality and enjoyment, particularly among participants who initially scored lower on a creativity measure.
The findings suggest that AI can help narrow certain differences in writing performance, giving less confident writers access to forms of assistance that might otherwise be unavailable. However, the experiment focused on short stories, not professional novels or journalism. Its findings should not be interpreted as proof that AI automatically improves every kind of writing.
Conquering the Blank Page
For many writers, the most difficult part of the creative process is beginning. The blank page represents an intimidating combination of possibilities and expectations. An author must generate ideas while simultaneously worrying about structure, vocabulary, coherence and style. Artificial intelligence can help separate these tasks.
A novelist might ask for several possible directions for a difficult scene. A journalist could explore competing approaches to an article’s introduction. A student might request questions that expose weaknesses in an argument. Most suggestions may be predictable or unsuitable. But even an unsuccessful suggestion can help a writer discover what the story should not become. This is where AI functions most effectively as a creative sandbox: a place to test possibilities without surrendering the final decision. The crucial distinction is between using AI to stimulate thinking and allowing it to substitute for thinking.
Shifting Attention to More Valuable Work
Writing involves many activities that are necessary but not especially creative. Authors must correct grammatical errors, reorganize paragraphs, eliminate repetition, check consistency and adapt material to different formats. AI can assist with these mechanical and editorial tasks, potentially leaving more time for research, reflection and original composition. It can also serve as a preliminary critical reader, identifying abrupt transitions, weak arguments or inconsistencies in a narrative. Yet these capabilities require supervision. AI may suggest corrections that change the intended meaning, overlook factual errors or replace an author’s distinctive language with generic expressions.
A writer who accepts every recommendation risks producing technically polished prose that has lost its personality. The objective should not be to eliminate the writer’s work but to concentrate that work where human insight matters most.
The Paradox of Artificial Creativity
Perhaps the most fascinating discovery about AI-assisted writing is that it can improve individual results while reducing originality across a larger group. The 2024 Science Advances study found that stories written with access to AI-generated ideas were generally evaluated more favorably, especially when their authors had lower initial creativity scores. But the researchers also discovered a troubling pattern: AI-assisted stories tended to resemble one another more closely than stories written without such assistance. This creates a paradox. A writer may produce a stronger individual story with AI, while widespread dependence on similar systems may gradually make published stories less diverse.
The explanation is not difficult to imagine. When thousands of writers consult systems trained on overlapping bodies of material, they may receive similar narrative suggestions, familiar metaphors and predictable structures. The resulting prose can appear sophisticated while expressing ideas that have become increasingly conventional. AI may therefore democratize access to polished writing while simultaneously encouraging a certain uniformity of expression. For literature and journalism, this is a serious concern. The value of writing lies not simply in communicating clearly but also in revealing perspectives that readers have not encountered before.
Who Owns the Words?
The growing use of artificial intelligence also raises difficult questions about copyright. Under current U.S. copyright principles, protection depends on human authorship. Material generated entirely by an AI system, without sufficient human creative contribution, generally does not qualify for copyright protection. However, using AI does not automatically deprive a writer of copyright.
The U.S. Copyright Office has explained that human-created elements of a work may remain protected when AI is used as an assisting tool. Creative selection, arrangement or modification may also qualify for protection, depending on the circumstances. The distinction is important. An author who uses AI to correct grammar is in a different position from someone who asks a machine to generate an entire manuscript and publishes it with minimal human involvement. Copyright protection is also not the same as permission to publish. Writers must consider contracts, the rights of third parties and the rules established by individual publishers.
What Publishers Expect
Publishing standards are evolving as artificial intelligence becomes more common. Some publishers and literary agents restrict AI-generated submissions or require disclosure of particular uses. Others distinguish between assistance with editing and the generation of substantial original content. Amazon Kindle Direct Publishing, for example, requires disclosure of AI-generated text, images and translations, but does not require authors to disclose ordinary AI-assisted editing, refinement or brainstorming. These distinctions demonstrate why writers should review the policies of the organizations with which they work rather than assume that every form of AI assistance is treated identically. Transparency is increasingly important, but so is precision about what the technology actually contributed.
Setting Boundaries: The Human Author Remains in Charge
The most productive relationship between writers and AI begins with clear boundaries. An author can use the technology to evaluate a manuscript’s structure without allowing it to rewrite the narrative voice. A journalist can request alternative interview questions while independently verifying facts and quotations. A student can ask for criticism of an argument rather than a completed essay.
Writers should also exercise caution when sharing unpublished manuscripts, confidential research or sensitive personal information with AI services. Privacy policies and data-handling practices vary. Perhaps the most useful principle is to treat AI as an assistant whose work must be questioned, not as an authority whose suggestions deserve automatic acceptance.
That means checking factual claims, rejecting clichés, preserving original observations and making conscious decisions about every substantial change.There is no universal rule requiring writers to rewrite an AI suggestion a particular number of times. What matters is whether the final work reflects genuine human creative judgment.
A New Definition of Writing?
Artificial intelligence is unlikely to settle the centuries-old question of what makes writing truly original. If anything, it makes that question more urgent. The technology can generate sentences, propose plots, organize arguments and imitate recognizable styles. But the existence of those capabilities does not eliminate the need for writers.
Instead, it places greater emphasis on qualities that have always distinguished meaningful writing: curiosity, imagination, experience, discernment and the courage to reject conventional answers. The future of writing may belong neither to authors who refuse artificial intelligence nor to those who surrender their work to it. It may belong to writers who learn to use the technology without allowing it to determine what they have to say. AI can help us find the words. The responsibility for having something worth saying remains ours.
