Gains:
- Being able to distinguish, on a task basis, at which stages of the writing process (brainstorming, draft, proofreading, summary) artificial intelligence saves real time, and at which stages (original idea, voice, final decision, fact verification) the responsibility remains with the author.
- Ability to apply a writing discipline that filters each AI output through the steps of linking it to the source, verifying it, and rewriting it in one's own voice.
- Understanding why creative voice, originality and copyright responsibility belong to the author and acquiring the habit of confidentiality and transparency
Behind every text is a silent series of decisions. Does this sentence or this flow better? Why is this character standing here? Is this section unnecessary? Which source supports this claim? Does this paragraph belong to my voice or is it something anyone can write? Writing and editing are the sum of thousands of these small decisions. Artificial intelligence (AI for short - computer systems that can produce text like humans, understand language, summarize and recognize patterns) sits right in the middle of these decisions. When used correctly, it reduces brainstorming, drafting, and proofreading from hours to minutes; When used incorrectly, it produces a text that erases your voice, introduces a fabricated fact into your text, or is too similar to another work.
This unit is not a software demonstration. Its purpose is to clarify where to put AI in your writing process and where not to put it at all. Let's lay out the basic principle from the beginning: AI is a writing assistant, not a writer. Responsibility for the original idea, creative voice, final judgment and fact-checking always rests with the author. Just as an architect uses a calculator but does not transfer the responsibility of the building to it, the writer uses AI but does not transfer the responsibility of the text to it.
Layers of the writing process and the place of AI
It is useful to divide the writing task into three layers. The creative layer is where the original idea, premise (the basic idea of a text, the "what if" question), voice, and theme are born; This is the core of the writer. The structural layer is planning, skeleton, partitioning and organization. The technical layer is grammar, spelling, consistency, style and proofreading. AI touches all three layers, but with different authority. At the technical layer, AI is a very powerful helper; generates quick drafts and options in the structural layer; At the creative layer, it only offers sparks and alternatives, the author makes the decision.
Let's define a few basic terms from the beginning. The draft is the first, raw text to be worked on. Proofreading is reviewing and correcting a text in terms of language, expression and consistency. Copyediting is the correction of grammar, spelling, punctuation and consistency. Synopsis is a text that summarizes the plot of a work. Localization is the adaptation of a text to another language and culture. We will explain these concepts one by one in the following units; For now, know this: In all of these, the AI gives you outlines and options, but not authorship decisions.
The following table summarizes the AI's role and author's share of responsibility by task:
Quest
Role of AI
Author's share
Risk
Brainstorming, list of ideas
spark generator
selection, development
low
Content plan / skeleton
sketch generator
Don't own the argument
low
First draft writing
Raw text generator
Process your voice
medium
Grammar/spelling correction
technical auditor
Meaning protection
low
Summary/synopsis
compressor
Check accuracy
medium
Fact/sourcing
Beginner tip only
Link to primary source
high
Original idea, sound, theme
Just suggestion
final decision
very high
Keep this one line in mind: as creativity and accuracy increase, the AI's role becomes smaller and the author's share grows.
Why the author's voice is inalienable
The author's voice is that recognizable trace that reveals who wrote a text: word choice, sentence rhythm, humor, spaces, repetitions, oddities. AI, by its nature, gravitates towards the most statistically probable, that is, the most generally valid expression. Therefore, when left to its own devices, it produces a plain, smooth but anonymous text; A language that anyone can write but belongs to no one. The reader recognizes an author from his voice and becomes attached to his voice. When you hand over your voice to AI, you hand over what makes you readable. So the rule we will repeat in every unit of this module is this: every text coming from the AI must go through your voice before it is published.
Why verification is the heart of this business
The AI seems confident in its answer, but it may not be sure. In technical language, this is called hallucination: it is the model's fabrication of information, source or quote that does not actually exist, in a fluent sentence, as if it were true. For a writer, this is a serious trap. The model can generate for you a non-existent quote such as "Tolstoy said about this"; It may show a study that does not exist as "a study conducted in 2018"; may give a date, a name, or a place name incorrectly. Since he says them all with the same fluency, the only thing that distinguishes right from wrong is your knowledge and your habit of verifying. The fabricated fact in a published text bears your signature; "AI wrote it" is not an excuse.
The discipline of verification for authorship is three steps:
- Link to source: For every fact, date, quote, name, and statistic, go to the primary source (book, magazine, official record, one-on-one interview), not the AI's memory. Use AI to interpret the source, not to remember it.
- Rewrite in your own voice: Don't leave every paragraph from the AI as is; rework it with your own words, rhythm and preferences.
- Test with the whole: Check from the author's perspective whether the output conflicts with the tone, theme, and intention of the text.
Attention: Publishing a fact, quote or source produced by AI without verifying it is like publishing a rumor whose source you have not seen as news. Just because the output is fluent is not true.
Privacy and copyright: your writing is your property
An unpublished manuscript, a script, a client text is your (or your client's) intellectual property. Loading these texts as is into a public AI tool poses the risk of the text being stored or used in model training, depending on the conditions of the tool. The rule is simple: don't share unnecessary, choose safe tool for sensitive text. If possible, choose corporate tools that have a data processing agreement and do not use your data in education; Share only anonymous snippets or structural questions in a top-secret project. On the copyright side, remember: AI output may resemble other works, and its legal protection may be unclear by country; Transforming it with your original contribution provides both ethical and legal assurance. We will deepen this topic in unit 11.
Basic templates you can use in every session
The following three templates will come in handy again and again throughout this module. The first is a "rule line" that reduces the risk of hallucinations and voice loss when you add it to the beginning of each AI session:
Rules (valid throughout this session):- DO NOT FABRICATE fact, date, source or quote; State where you are unsure as "not sure". - Do not produce numbers or names without me giving data/context. - Stay true to the tone and claim I give; shifting to your own general tone.- Briefly list each important assumption you made at the end.
The second is for an anonymization check before sharing sensitive text:
Task: Is there personal/confidential information (name, institution, address, private details) in the text below? If so, list them and suggest an anonymous response for each (e.g. "Ahmet Yıldız" -> "a middle-aged teacher"). I will clear these before sharing the text. Just detect; publish or store text.
The third is a quick control template for pulling an output into your own voice:
Below is an outline. REWRITING.Task: Mark commonplace sentences and cliché expressions that "anyone can write" in this text. Give a one-word warning for each (generic / cliché / empty). I will make the correction with my own voice.
Tip: Keep these three templates in a notes file. Starting each new writing session with AI with a rule line prevents most of the mistakes you'll notice later in the first place.
three mini cases
Case 1 — Safe use. An essayist was having difficulty creating the skeleton of a 1,500-word article. He gave AI its subject, target reader and main claim and asked for three different skeletons. The AI generated three options in 2 minutes; the author chose one and filled it with his own arguments, checked each fact from its source and wrote the text in his own voice. Planning time reduced from 90 minutes to 20 minutes; The text was entirely the author's.
Case 2 — Made-up quote trap. One blogger asked AI to “give a famous quote about this.” AI made a promise that never existed by attributing it to a famous name. The author published it without verification; readers found the quote to be fabricated, and the author's credibility was damaged. Mistake: using the quote without verifying it in the primary source.
Case 3 — Loss of voice. A novelist had an entire chapter written by an AI and put it into his book as is. The editor noticed that that section looked "as if someone else had written it" and that it was disconnected from the voice of the book. The text was smooth but anonymous. The right way: to take the AI draft as a skeleton and rewrite it in the author's own voice.
Weak prompt / Strong prompt
Weak prompt:
Write me a good essay on loneliness.
This prompt is flawed: there is no target reader, no length, no tone, no point of view, and no opinion of the author himself. AI only produces a generic, anonymous text.
Powerful prompt:
Your role: draft assistant assisting an essayist. Subject: the two faces of loneliness in modern city life (liberating / exhausting). Target reader: literary magazine reader. Length: 1,200 words. Tone: introverted, reflective. My main claim: loneliness is not a deficiency, but an area that can be chosen. Task: Give a 4-chapter skeleton and 2 sentences of direction for each chapter. DO NOT make up a fact, quote or name; Mark "verify" where you are not sure.
Here the role, context, claim, length, tone and prohibition of "making up" are clear. The output is a start; You still do the writing and verification.
Common mistakes
- Waiting for summer before giving your own opinion to the AI. If the idea does not come from you, the text will be generic and unfamiliar to you.
- Mistaking fluent output as correct. A well-written sentence does not mean it is factually correct.
- Delegating your voice. Leaving the AI's flat tone intact is erasing the trace that makes you readable.
- Not verifying the source/citation. A fabricated source becomes an error published with your signature.
- Loading the hidden text into the open tool. Unpublished work is intellectual property; Unnecessary sharing is a risk.
Tip: Include a short “rule line” in every AI session: “Do not make up facts, sources, or quotes; state clearly where you are unsure; stick to the tone and argument I give.” This single sentence significantly reduces the risk of both hallucinations and voice loss.
In summary
Artificial intelligence is a powerful assistant in writing and editing: it speeds up brainstorming, drafting and proofreading. But the original idea, creative voice, final judgment and fact-checking always belong to the author. As creativity and accuracy increase, the role of AI becomes smaller. Three habits are the basis of everything: linking to the source, rewriting in one's own voice, and not sharing the unnecessary. Once you internalize these three, every technique in the rest of the module becomes a safe accelerator for you.
Application task
Take a small writing idea you have or have dreamed of (an essay, a story opening, a blog post). Using the "Strong prompt" template above, ask the AI for just one skeleton. Then: (1) fill in the skeleton with your own argument, (2) check each fact/quote from its source if applicable, (3) write one paragraph from AI's draft, one paragraph in your own voice, and compare the two. Note the difference in 5 items: which sentences are "yours" and which are "everyone's"?
checklist
- [ ] Have I given the AI my own opinion, tone, and context?
- [ ] Have I verified every fact, date, and quote in the printout from the primary source?
- [ ] Did I rewrite the text using my own voice?
- [ ] Have I added the "fabricate fact/source" rule to the prompt?
- [ ] Have I moderated safe tools and unnecessary sharing of sensitive or unpublished text?
- [ ] Did I retain final authorial judgment and responsibility?