Direct answer
No — but it will change where creativity lives. Controlled studies show that generative AI raises the judged creativity of an individual piece of work while simultaneously making everyone's work more alike, so the real risk is not the death of human creativity but its quiet homogenization. The writers who keep their originality are the ones who use AI as a sparring partner for ideas and keep the drafting, judging, and revising in their own hands [1].
Does AI Actually Make Human Creative Output More Uniform?
The strongest evidence says yes, at the population level, even when individual quality improves. In a large experiment on short-story writing, researchers gave some writers access to generative AI idea prompts and then had the resulting stories scored for novelty, usefulness, and emotional appeal. AI-assisted writers scored higher on every dimension — but the stories drifted toward each other, and the collective diversity of the output fell sharply [1]. The mechanism is straightforward: a model trained on a shared corpus proposes suggestions drawn from a shared distribution, so millions of writers anchoring on those suggestions converge on the same metaphors, structures, and phrasings [2].
That convergence is measurable at the corpus level rather than the document level, which is why it is easy to miss from the inside. A single essay can feel fresh and distinctive to the student who wrote it while still sitting squarely inside the modal style of every other essay produced that week with the same tool [2]. This is the crucial distinction the "AI will kill creativity" argument usually gets wrong: homogenization is a collective risk, not an individual one, and it accumulates through millions of small, individually reasonable choices to accept a suggestion.
There is also a distributional twist that complicates the pessimism. The creativity gains from AI assistance are largest for writers who initially score lowest on creativity and smallest — sometimes negative — for those already at the top [1]. In other words, AI compresses the range. It is a levelling technology, and levelling is good news for the median writer and bad news for the cultural value of the outlier, which is exactly the part of human creativity that history tends to remember [2].
Does Outsourcing Thinking to AI Weaken the Writer's Own Cognitive Skills?
Self-reported evidence from knowledge workers points to a real erosion of effort, and the pattern is counterintuitive. Survey research on generative AI use found that people who trusted AI output more reported less critical thinking, not more — confidence in the tool substituted for scrutiny of it [3]. Participants described shifting their effort from generating content to verifying it, and in many cases the verification step was thin or skipped entirely [3]. The skill that atrophies first is not vocabulary or grammar; it is the willingness to disagree with a fluent paragraph.
This matters for creativity specifically because creative work is built out of the friction that offloading removes. Deciding which of five bad openings to keep, noticing that an argument does not actually hold, and rewriting a sentence you are proud of are the exercises that build a writer's judgment. When those decisions are pre-made by a model, the finished product may look fine while the underlying capability quietly stops developing [3]. Cognitive offloading is not automatically harmful — a calculator does not destroy arithmetic ability in people who still learn the algorithm first — but it is harmful when it replaces the practice rather than the tedium.
The practical warning sign is confidence without provenance. If you cannot explain why a sentence is in your essay, or reconstruct the reasoning that led to your conclusion, the AI did the creative work and you did the transcription [3]. That is the honest version of the "AI kills creativity" fear, and it is a claim about habits rather than about machines.
How Can a Student Keep Creative Ownership While Still Using AI?
Ownership is preserved by sequencing, not by abstinence. Generate your own ideas and a rough structure before you open a chat window, then use AI to stress-test and expand what you already have rather than to originate it. Because AI's creativity benefit is largest when it supplies starting points you would not have reached alone [1], the goal is to take the divergent suggestion and then deliberately diverge again — rewrite it in your own register, add a detail only you could know, and cut anything that sounds like the model's default voice [2].
Measurement is the second half of the method, because you cannot correct what you cannot see. Turnitin's AI Writing Report shows the percentage of qualifying text flagged as AI-generated and highlights the specific sentences responsible, which turns a vague worry into a list of passages you can actually rewrite [4]. It is worth knowing how to read that output: scores below the 20% confidence threshold display as an asterisk rather than a number, and the report is designed as a signal for a conversation about authorship rather than proof of misconduct [4]. Used that way, it is a diagnostic for where your own voice stopped and the model's began.
Finally, treat the flagged passages as a revision syllabus rather than a verdict. The sentences a detector lights up are usually the ones with the least specific detail, the most generic transitions, and the fewest commitments — which is precisely the writing that homogenization produces [2]. Rewriting them by hand, with a concrete example, a named source, or a claim you are willing to defend, does two things at once: it lowers the AI signal and it rebuilds the judgment muscle that offloading had been resting [3]. That is how you keep the tool without renting out the author.
The argument above only works if you can see your own draft clearly — which is hard to do when you have been staring at it for a week. That is the gap turnitin0 exists to close: a pre-submission check that returns the same Turnitin AI detection report and similarity report your professor will see, so the passages where your voice gave way to the model's are named, highlighted, and fixable before the deadline rather than after it.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Does using AI for brainstorming count as cheating?
Usually no, and the distinction most institutions draw is between generating ideas and generating submitted text. Brainstorming with a model is closer to talking to a tutor or a peer, whereas submitting its prose as yours is the part that raises authorship questions [4]. Check your own course policy first, since the line varies by institution.
If AI makes everyone's writing more similar, does that mean my AI-assisted essay is bad?
Not automatically — individual quality and collective diversity are separate measures, and AI assistance tends to raise the first while lowering the second [1]. The essay can be competent and still sound like everyone else's. The fix is deliberate divergence: specifics, personal evidence, and a voice the model would not have chosen [2].
Can I rebuild creative skill if I have already been leaning on AI?
Yes, and the recovery is unglamorous: write first drafts unaided, argue with the model instead of accepting it, and verify claims rather than trusting fluency [3]. Effort is the input that builds judgment, so reintroducing friction is the mechanism, not a punishment.
Why does my Turnitin AI score show an asterisk instead of a number?
Turnitin displays an asterisk when the AI score falls below its 20% confidence threshold, meaning the signal is too weak to report as a precise percentage [4]. It indicates low-confidence detection rather than a guaranteed clean result, so read the highlighted passages alongside the score.
Will AI eventually replace human creativity entirely?
The evidence supports substitution of tasks, not of the creative faculty itself, because the homogenization effect depends on people choosing the model's default [1][2]. As long as some writers deliberately diverge, the space for original work persists — and the writers who diverge are the ones who stay legible as authors [3].