In January 2024, Japanese writer Rie Kudan won the Akutagawa Prize for Tokyo-to Dojo-to. At the press conference she dropped a single sentence that detonated across the literary world: roughly 5% of the text came directly from ChatGPT.
The judges went back to the manuscript. They compared passage by passage. Not one of them could identify which 5%.
The AI-written sentences and the human-written sentences had merged -- completely indistinguishable.
Nobody said the novel was bad. The conversation shifted to something stranger: if nobody can tell the difference, what exactly is the argument about?
The Essence of "Remixing"
The word "plagiarism" gets thrown around. It lands in the wrong place.
Large language models train on massive deposits of human language -- millions of books, countless articles, the entire sedimented output of the internet. When AI produces a passage, it rearranges and recombines patterns it has encountered before.
But human creation does the same thing.
T.S. Eliot said it bluntly enough: "Immature poets imitate; mature poets steal." Murakami's voice traces back to Raymond Carver. Carver's to Hemingway. Hemingway's to Gertrude Stein. Every writer stands on the shoulders of predecessors -- or more precisely, on their corpus.
The difference might lie in consciousness. A human writer knows what she is doing: selecting, rejecting, warping influence. AI has no such self-awareness. It runs statistical computation.
But if the final output cannot be distinguished, does that "consciousness during the process" actually change anything?
No standard answer exists. The question simply tears open the core fracture of this debate -- do we care about the "process" of creation, or the "result"?
The Question of Humanity
Literature has never existed for the sake of beautiful sentences.
It exists because one mind touches another. Fear, hope, confusion, epiphany -- these things travel through text, leaping from the author's head into the reader's. That connection is what makes someone cry over a novel, laugh at a passage, stare at the ceiling at 3 a.m. thinking about a fictional character's fate.
AI has no experiences. No fear. No memory of sleepless nights. When it writes "she felt a wave of sadness," it has no idea what sadness feels like as a texture.
The critic's challenge is powerful: if the author is not a real mind, why read?
Except -- return to Rie Kudan. The most elite literary readers in Japan sat on that judging panel, and none of them detected a trace of "non-human." At minimum, in certain moments, AI-produced text does evoke enough resonance for readers to believe they are touching another mind.
Maybe the question should face the other direction. A painting does not require the painter to actually feel sadness while painting in order to make the viewer weep. A piece of music does not require the composer to have experienced love while composing in order to shatter the listener's heart.
The emotional effect of a creative work and the emotional state of its creator do not necessarily need to match.
This is not a defense of AI. It is an uncomfortable fact: our judgment of "humanity" may depend more on the text itself than on the identity of the author.
The Evolution of Tools
Socrates hated writing.
Not an exaggeration -- Plato's Phaedrus records the concern in black and white: once people start writing things down, they will stop truly remembering anything. Writing would destroy memory.
When the typewriter arrived, critics said the intimacy of handwriting was lost -- this was not "real writing." When word processors appeared, critics said the delete key made writing too cheap -- revision no longer carried a cost.
Every time, the fear was real. Every time, people adapted.
Spell check is now standard equipment. Grammar suggestions appear everywhere. Auto-complete lives inside every writing application. All of these are forms of AI assistance -- we just stopped calling them that.
The supporters' core argument sits here: true creation happens in the conception phase, not the typing phase.
Who decides what this story will say? Who designs characters, plans plot arcs, determines theme? Who judges which draft is worth keeping and which should be discarded? Who bears the final creative responsibility?
The person writing.
AI is an instrument of execution, not the source of creativity. Just as the brush was never the artist.
Inside Slima's Writing Studio, this distinction becomes especially visible when working with the AI Assistant. The AI might generate five versions of a dialogue exchange -- but which to choose, how to modify it, why one version works better than another -- those judgments belong entirely to the creator. The tool amplifies capability. The steering wheel stays in human hands.
The Gray Zone
Reality offers no clean boundaries. That is precisely why this debate burns so fiercely.
Picture a spectrum. Far left end: alone at a desk, pen on paper, every word written by hand. Far right end: typing "write me a fantasy novel," publishing whatever the AI produces.
Both extremes are easy to judge. One raises no concerns. The other raises plenty.
The vast middle territory is where things get complicated.
Asking AI for five possible plot twists, then picking one and developing it personally -- does that count as AI writing?
Writing a dialogue scene, feeling stuck, feeding it to AI for polishing, adopting two of its suggested lines -- does that count as AI writing?
Using AI to generate an entire first draft, then spending three months gutting and rebuilding it, injecting personal style and experience throughout -- does that count as AI writing?
Where is the line? Who has the authority to draw it?
No standard answers exist. But these questions expose something critical: "AI writing" is not a binary classification. It is a continuous spectrum.
Version Control makes this spectrum traceable. Every revision, every Snapshot, every experimental branch records the human creator's trail of judgments. AI provides material. Humans provide choices -- and the history of those choices is the evidence of creation.
A Framework for Thinking
Rather than getting tangled in the too-large question of "is AI writing good or bad," it may help to break it into smaller, more manageable pieces.
On the nature of creation. Does value lie in process or result? If equally good work emerges from less effort, is that laziness or efficiency? An oil painter does not lose the title of "artist" for switching to acrylic -- but where is the boundary?
On authorship. When AI participates in part of the text output, does the label "author" still hold? If the story's conception, judgment, and revision all come from the human, but not every word was typed by the human -- how is this fundamentally different from hiring an editor who makes sweeping revisions?
On the reader's right to know. Do readers deserve to know whether a book used AI? Rie Kudan volunteered the information -- but if she had stayed silent, could anyone have pressed the claim? What role should transparency play here?
On growth. This is the most personal question. Does using AI improve writing ability or erode it? Like relying on a calculator until mental arithmetic atrophies -- could AI cause the writing muscle to weaken?
The design logic of AI Beta Readers responds directly to this last concern. They do not write for the creator. They provide feedback from the reader's perspective -- like a mirror reflecting parts of the work the creator cannot see on their own. Growth comes from confronting feedback, not from outsourcing the writing.
Our Position
As the team building an AI writing tool, avoiding a stance would be dishonest.
AI is a thought amplifier, not a creativity substitute.
What does that sentence mean when unpacked?
AI can accelerate the distance between an idea and the words on the page. It can flag blind spots, offer angles not previously considered, handle mechanical tedium -- format conversion, grammar correction, multilingual translation.
But it cannot decide what a story should say. It does not know what matters to a specific readership. It lacks the irreplaceable sediment of a writer's lived experience. It cannot bear responsibility for a creative decision.
The most effective AI collaboration: AI handles "how to say it," the human focuses on "what to say" and "why to say it."
This is not a diplomatic platitude. It is an observation drawn from thousands of use cases inside Writing Studio: people who treat AI as a ghostwriter produce work that all looks the same -- smooth, correct, devoid of personality. People who treat AI as a conversation partner produce work that becomes sharper, because the energy saved gets redirected to what genuinely matters.
What Does the Future Hold?
Technology only runs forward.
AI-generated text will become increasingly difficult to distinguish from human writing. "AI detectors" will see their accuracy decline steadily until, at some point, detection itself becomes meaningless. Services claiming to identify AI text are already producing waves of false positives.
Norms will form slowly -- but slowly. Photography took nearly a century to gain formal recognition as an independent art form. AI-assisted writing will likely take no less time for society to digest.
A paradoxical trend is surfacing: as AI content floods the market, "purely human creation" may become a scarce resource.
The Industrial Revolution turned handmade goods into luxury items. In the age of mass production, handcraft actually became more expensive. The same logic may apply to the writing market -- one track for AI-assisted high-efficiency content production, another for the "artisanal goods" of purely human creation. The scarcity of the latter could make it more collectible.
The two tracks do not conflict. They serve different needs.
Final Words
Is AI writing plagiarism?
This question has no clean "yes" or "no." It entangles the definition of creation, the relationship between tools and humans, the boundaries of art, society's expectations of originality. These debates have raged for thousands of years -- they merely have a new vessel now.
But one thing requires no debate --
What ultimately gets remembered is the work itself. Whether it has value, whether it is sincere, whether it can shift -- even slightly -- the way another person sees the world after reading it.
Tools will keep evolving. Fear will persist. But good stories -- regardless of how they were created -- will always find readers.
The remaining questions? Every writer must answer them alone.