• Home
  • The Company
  • Services
  • How It Works
  • Pricing
  • Blog
  • Contact
  • Suite 3AB, 4th Floor, No 34/36 Ikorodu Road, Jibowu, Lagos
  • +2349132683414
  • [email protected]
  • Home
  • The Company

      Our Agency

      Writely is a leading content writing agency in Nigeria, delivering tailored solutions for businesses and individuals.

      Learn More

      Our Values

      To provide professional content and writing services in Nigeria, creating impactful content that resonates with audiences and drives success.

      Learn More

      Our Vision and Mission

      We aim to be the go-to content writing agency in Nigeria, empowering every story to inspire and connect.

      Learn More

      FAQs

      Find answers to questions about our content writing services and processes in Nigeria.

      Learn More

  • Services
  • How It Works
  • Pricing
  • Blog
  • Contact
Free Consultant

  • Home
  • The Company

      Our Agency

      Writely is a leading content writing agency in Nigeria, delivering tailored solutions for businesses and individuals.

      Learn More

      Our Values

      To provide professional content and writing services in Nigeria, creating impactful content that resonates with audiences and drives success.

      Learn More

      Our Vision and Mission

      We aim to be the go-to content writing agency in Nigeria, empowering every story to inspire and connect.

      Learn More

      FAQs

      Find answers to questions about our content writing services and processes in Nigeria.

      Learn More

  • Services
  • How It Works
  • Pricing
  • Blog
  • Contact
Free Consultant

Content Writing

Does Turnitin Detect AI Writing? What Students Should Know

By Splendor 

Yes, Turnitin can detect AI writing, but only where your university has switched the feature on, and the result is an estimate, not proof. That one sentence answers the question, but it leaves out almost everything students actually need to know. Students are worried about two opposite things at the same time: getting caught for using AI tools where they are not allowed, and being wrongly accused when they wrote the work themselves. Both worries are reasonable, and both can be handled better with accurate information.

This guide explains what Turnitin’s AI writing indicator is, how it differs from the similarity score, what its numbers and the asterisk mean, how reliable it is, what universities do when work is flagged and what to do if you are the one flagged. It also covers how to use AI tools responsibly where your institution allows it, and why shortcuts such as ‘humaniser’ tools are a bad idea. It is written for Nigerian students working on assignments, projects, dissertations and theses, but the principles apply anywhere Turnitin is used.

If you want an expert to review your draft for weak paraphrasing, missing citations and passages that do not sound like you, see our AI content detection and revision service. You can also browse all of our Citation and Referencing Services or start from the Writely.ng homepage.

Table of Contents

Toggle
  • The Quick Answer
  • Two Different Tools: Similarity Score vs AI Writing Indicator
  • Does Your University Use the AI Indicator?
    • Questions to ask your department
  • How Turnitin’s AI Writing Detection Works (in Plain Language)
  • What Counts as Qualifying Text?
  • Reading the AI Report: Numbers, Asterisk and Colours
    • What the percentage does and does not mean
    • Three short examples
  • How Reliable Is Turnitin’s AI Detection?
    • What Turnitin says about accuracy
    • False positives: human writing flagged as AI
    • False negatives: AI text that is missed
    • Models change, and results can change with them
    • What this means for you
  • Can Turnitin Detect Paraphrased, Edited or ‘Humanised’ AI Text?
  • What Happens When Work Is Flagged?
    • What supervisors and panels look at besides the score
  • If You Used AI Tools: Rules, Disclosure and Good Practice
    • A sensible routine
    • An illustrative disclosure statement
    • Referencing AI tools
  • If Your Own Writing Is Flagged
    • Step 1: Stay calm and do not delete anything
    • Step 2: Understand exactly what was flagged
    • Step 3: Collect your evidence
    • Step 4: Prepare to explain your process
    • Step 5: Be ready to discuss the content
    • Step 6: Talk to your supervisor early
    • Step 7: Know the process and your rights
    • Step 8: Learn from it
  • How to Protect Yourself: Build a Visible Writing Process
  • Writing Habits That Keep Your Own Voice Visible
  • Humanisers, Spinners and ‘AI Bypass’ Tools: Why to Avoid Them
  • Students Who Write English as a Second Language
  • AI and Your Dissertation: Chapter-by-Chapter Integrity
  • Preparing for the Viva
  • Myths vs Reality
  • Common Mistakes Students Make About AI Detection
  • AI Writing Readiness Checklist
  • Frequently Asked Questions
  • Related Reading Across the Cluster and the Dissertation Series
  • Need Help Reviewing Your Draft?

The Quick Answer

  • Turnitin has a separate AI writing indicator. It is different from the similarity score that most students know. See What Is a Good Turnitin Similarity Score?
  • It only works where your institution has it enabled. Some universities use it, some restrict who can see it and some have switched it off.
  • It gives an estimate, not a verdict. It reports the share of qualifying text that looks like AI-generated writing. Turnitin itself says the result should not be the only basis for an academic misconduct decision.
  • Scores from 1% to 19% appear as an asterisk. Turnitin treats results below 20% as lower confidence, so it shows *% and does not highlight text.
  • Errors are possible in both directions. Human writing can be flagged (false positives), and AI text that has been rewritten or mixed with human writing can be missed (false negatives).
  • A flag should start a conversation. Supervisors and integrity panels usually look at your drafts, your notes, your earlier work and whether you can explain what you wrote.
  • Your best protection is a visible writing process. Keep dated drafts, notes and source logs, write from understanding and know your university’s AI policy.

Two Different Tools: Similarity Score vs AI Writing Indicator

Most students know Turnitin as a plagiarism checker. That tool is the similarity report. The AI writing indicator is a different tool, and the two should not be confused.

FeatureSimilarity reportAI writing indicator
What it doesMatches your text against web pages, journals, books and previous student papersAnalyses the writing itself and estimates how much looks like AI-generated text
What it reportsA percentage of matching text, with sourcesA percentage of qualifying text likely to be AI-generated, with highlights above a threshold
Needs a source to match?Yes. It needs the same text to exist elsewhereNo. It does not look for a matching source
Where highlights appearIn the similarity reportIn a separate AI writing report
Can a person explain a match?Yes, because you can compare your text with the sourceNot in the same way, because there is no source to compare
Affected by citations and quotation marks?Yes, quotations and references can matchNo, the tool looks at the style of the writing, not at citations

This difference matters in practice. A low similarity percentage says nothing about AI use. A student can write an entire chapter with an AI tool, and it may produce very few similarity matches because the text is newly generated. In the other direction, a student can have a high similarity percentage from properly quoted material and a clean AI result. If your university uses both tools, you may receive two separate results, and you should read them separately.

For help reading the similarity report, see What Is a Good Turnitin Similarity Score? and How to Lower Your Plagiarism Percentage Ethically.

Does Your University Use the AI Indicator?

Not every university does. The AI writing indicator is a feature that an institution licenses and switches on, and policies about who can see it vary widely. In practice, you may find one of these situations:

SituationWhat it means for you
Enabled for staff onlyInstructors can see the AI result, but students cannot. You may not know your score unless you are told
Enabled for staff and studentsYou may see the AI result on your own submission report
Not enabledYour submissions get a similarity report only
Enabled for some modules or departmentsPolicies differ between courses at the same university
Used only when there is a concernThe tool is run on request, for example when a supervisor suspects a problem

Some universities worldwide have decided to switch the AI indicator off because of concerns about false positives. Others rely on it as one piece of evidence among several. There is no single Nigerian or international rule, so the only reliable way to know is to ask.

Questions to ask your department

  1. Does our university use Turnitin’s AI writing indicator, and is it switched on for my course?
  2. Can students see the AI result on their own submissions?
  3. What is the university’s policy on AI tools: banned, allowed with disclosure or allowed for specific tasks?
  4. How do I disclose AI use if it is permitted?
  5. What happens if my work is flagged, and what is the process for explaining or appealing?
  6. Are drafts submitted for feedback checked by the AI indicator as well?

Write the answers down. If a policy is unclear, ask for it in writing, because a verbal reassurance is hard to prove later.

How Turnitin’s AI Writing Detection Works (in Plain Language)

Turnitin does not publish the full details of its model, so no outside guide, including this one, can describe it exactly. What Turnitin and independent commentators have said in public is enough to understand the basics.

1. The text is split into segments. The tool takes your submission and breaks the qualifying text into overlapping chunks of a few sentences.

2. Each segment is scored. A classifier looks at statistical patterns in the writing and estimates how likely each segment is to have been produced by a large language model. In broad terms, AI-generated text tends to be very smooth and predictable at the level of word choice and sentence flow, while human writing tends to be more varied and idiosyncratic. The classifier is trained on examples of both.

3. Sentences are labelled. Segments that cross the threshold are marked as likely AI-generated. Newer versions of the tool can also mark text that appears to have been generated by AI and then reworded with an AI paraphrasing tool, and they show these in a different colour.

4. An overall percentage is calculated. The tool reports the share of qualifying text that was flagged. The percentage is a share of the text, not a measure of how much effort or thought went into the work.

5. The result is shown with a confidence rule. Results below 20% are displayed as an asterisk rather than a number, because Turnitin treats them as less reliable.

Two consequences follow from this design.

First, the tool does not know which AI tool produced a passage, and it cannot prove that you used one. It estimates how much of the writing resembles patterns typical of AI text. It does not compare your essay with a stored copy of a chatbot’s answer.

Second, the tool looks at style, not at truth or source. It does not check whether your arguments are correct, whether your citations are real or whether you understood your reading. Those are things that supervisors and examiners check.

What Counts as Qualifying Text?

The indicator is designed for long-form prose, so it only evaluates what Turnitin calls qualifying text. This means paragraphs of complete sentences, such as the body of an essay, a literature review or a discussion section.

Text that is generally outside what it evaluates includes:

  • bullet points and lists,
  • tables and figure captions,
  • headings and titles,
  • code and mathematical expressions,
  • very short submissions. Turnitin documents a minimum amount of qualifying prose (stated as 300 words at the time of writing) before it will produce a score.

There are also limits on file type, size and language. Turnitin’s documentation changes over time, so check the current requirements at Turnitin’s guidance pages if you need exact figures.

A practical consequence: a submission that is mostly tables, lists and data may produce no AI score at all, while a literature review or discussion chapter made of long paragraphs is exactly the kind of text the tool evaluates.

Reading the AI Report: Numbers, Asterisk and Colours

If your university lets you see the AI result, this is how to read it.

What you seeWhat it means
No score, or a message that the document does not qualifyThe submission did not meet the requirements, for example it was too short or in an unsupported format
0%No qualifying text was flagged as likely AI-generated
*% (asterisk)Some text resembled AI writing, but the result was above 0% and below 20%. Turnitin treats this as low confidence, shows no number and does not highlight text
20% to 100%The percentage of qualifying text identified as likely AI-generated (or AI-paraphrased) is shown, and the relevant text is highlighted

Two features of this scale are worth understanding.

The 20% threshold. Turnitin introduced the asterisk in 2024 so that low scores would not be misread. Reports created before that date may still show a number below 20%. Turnitin’s own guidance says that low-confidence results should not be treated as evidence of misconduct, and the asterisk acts as a constant reminder of that.

The highlight colours. In many versions of the report, one colour marks text that appears to be AI-generated and a different colour marks text that appears to be AI-generated and then modified by an AI paraphrasing tool. The exact colours and labels depend on the version your institution uses, so read the key shown in your report.

What the percentage does and does not mean

  • It is a share of qualifying text. A score of 40% means that about two-fifths of the evaluated prose was flagged, not that 40% of your effort was automated.
  • It is not a measure of how much AI you used. You can use AI for ideas, outlines or feedback and produce no AI-like prose. Equally, a human can write prose that the classifier flags.
  • It is not a measure of plagiarism. AI text is not copied from any source, and a plagiarised passage is not necessarily AI-written.
  • A low score is not proof of innocence, and a high score is not proof of guilt. Either result is an input to a human judgement.

Three short examples

These scenarios are invented to show how to think about the numbers.

Example 1: An asterisk. A student’s discussion chapter shows *%. The student wrote it herself, and two sentences happen to be flagged because they are formal and formulaic. Because the result is below 20%, Turnitin shows no highlights and treats it as low confidence. There is no case to answer.

Example 2: A mid-range score. A student’s literature review shows 32%, with highlighted paragraphs that are all smooth, general summaries. The supervisor asks to see the student’s drafts and notes. The student shows dated outlines, annotated PDFs and a source log. The supervisor asks a few questions about the sources, and the student answers them well. The conversation ends with advice to add more critical analysis, and the flag is not pursued.

Example 3: A high score with no process. A student’s whole methodology and discussion show 85%, there are no drafts, the student cannot explain several choices and two references do not exist. The panel treats this as a serious concern, not because of the percentage alone, but because the other evidence points the same way.

The lesson is the same each time: the score opens a conversation, and the surrounding evidence decides what happens next.

How Reliable Is Turnitin’s AI Detection?

This is the question behind most of the anxiety. The honest answer is that it works reasonably well on some kinds of text, works poorly on others and should never be treated as proof by itself.

What Turnitin says about accuracy

Turnitin has stated publicly that it designs the tool to keep false positives low at document level for submissions that contain a large amount of AI writing, and it has given a target rate of under 1% for documents with more than 20% AI writing. It has also acknowledged that errors are more likely at sentence level and at low percentages, which is part of the reason for the asterisk and the 20% threshold. It advises that the indicator should be used as a starting point for a conversation and not as the sole basis for action.

These are the company’s own statements. Independent studies of AI detectors in general have found mixed results, with detection performing well on raw, unedited AI text and worse on text that has been edited, paraphrased or mixed with human writing. Treat all accuracy figures, from any source, as approximate and subject to change as models and tools evolve.

False positives: human writing flagged as AI

A false positive is when a human-written passage is flagged as AI-generated. Factors that can raise the risk include:

FactorWhy it can raise risk
Formal, formulaic writingStandard academic phrasing, definitions and boilerplate can look statistically predictable
Highly edited or polished draftsHeavy smoothing, whether by hand or with grammar tools, can reduce natural variation
Technical or definitional passagesThere are only so many ways to describe a standard procedure or define a term
Templates and model textWriting that follows a template can look uniform
Short textsLess text means less evidence, so scores are less stable
Writing by non-native English speakersSome researchers and universities have raised concerns that detectors may be less fair to writers working in a second language
List-like or repetitive structureRepetitive sentence patterns can look machine-like

The consequence is simple: a flagged passage is not necessarily AI-written, and students who write in a careful, formal style are not automatically safe.

False negatives: AI text that is missed

A false negative is when AI-assisted text is not flagged. This can happen when:

  • AI output has been substantially rewritten by a person,
  • AI text is mixed with the student’s own writing,
  • the AI text is short,
  • the AI tool or its output style is one the model handles less well,
  • detection models have not yet caught up with newer AI tools.

Turnitin has said that its detection of AI-paraphrased text is less reliable than its detection of raw AI output, and it has added features aimed at flagging text altered by paraphrasing tools. These features change over time, so do not assume that any pattern of editing is invisible or visible.

Models change, and results can change with them

Detection models are updated. A document that scored one way last year may score differently if it is resubmitted after a model update. This is another reason why no single result should be treated as final.

What this means for you

  1. Do not treat the indicator as a judge. It is one input.
  2. Do not assume you are safe because your score is low, or doomed because it is high.
  3. Do not try to game it. Evasion tools change meaning, can be detected and are treated as misconduct in many institutions.
  4. Build evidence of your own process, which is what protects you whatever the score says.

Can Turnitin Detect Paraphrased, Edited or ‘Humanised’ AI Text?

This is the question behind many searches, and it deserves a careful answer.

Type of textHow the detector tends to behave
Raw, unedited AI outputThe category the tool is built for. It is the most likely to be flagged
AI text lightly edited by a personOften still flagged, because the underlying patterns remain
AI text heavily rewritten by a personLess likely to be flagged, and the result is less reliable either way
AI text run through an AI paraphrasing toolTurnitin says it can flag this separately, but it describes detection as less reliable than for raw output
A human draft polished with a grammar or style toolUsually fine, but heavy automated rewriting can raise the risk of a flag
Human writing in a very formal or formulaic styleCan be flagged by mistake
A mixture of human and AI textResults vary, and the percentage reflects only the flagged share

Three points follow.

First, detection is not a fixed target. Turnitin and other vendors update their models, and they have said they are working on tools aimed at text altered by ‘bypasser’ or ‘humaniser’ programs. Something that is not flagged today may be flagged after the next update, and the reverse can also happen.

Second, evasion is not a safe strategy. Even if a tool slipped past a detector, the work would still have problems that matter more: changed meaning, awkward wording, invented references and a mismatch with your normal writing. Supervisors notice these. Many universities also treat deliberate attempts to disguise AI use as a separate and more serious offence.

Third, the real test is whether you can explain your work. A vague, generic paragraph that you cannot defend is a problem whatever a detector says. A well-argued paragraph in your own voice, which you can discuss in detail, is not.

What Happens When Work Is Flagged?

Procedures differ by institution, but a typical sequence looks like this:

  1. The result is noticed. An instructor or supervisor sees a high AI percentage, or a passage that looks unlike your usual writing.
  2. The instructor looks for more evidence. They may compare the work with your earlier submissions, check the highlighted passages, check whether the references exist and look at your drafts if they are available.
  3. You are asked for an explanation. This may be an informal conversation, a written statement or a meeting with an academic integrity officer.
  4. You are asked about the content. A common test is to ask you to explain specific passages, sources or methodological choices, sometimes in person.
  5. A decision is made. Possible outcomes range from no action, through feedback and a requirement to revise, to a formal misconduct finding, depending on the evidence and the institution’s rules.
  6. There is usually an appeal route. Check your student handbook for the process and the time limits.

Turnitin’s own guidance, and the policies of many universities, say that an AI indicator result should not be the only basis for a misconduct finding. In practice, the strongest cases combine several kinds of evidence, while the weakest rest on a percentage alone. Knowing this should reduce panic, but it should not make you careless.

What supervisors and panels look at besides the score

  • whether the writing style matches your earlier work,
  • the version history and dates of your drafts,
  • your notes, outlines and annotated sources,
  • whether the references exist and say what you claim they say,
  • whether you can explain and defend the content,
  • whether your explanation of how you worked is consistent and believable.

If You Used AI Tools: Rules, Disclosure and Good Practice

University policies on AI tools vary a great deal, and they are changing quickly. You may find one of these approaches at your institution:

Policy typeWhat it usually means
ProhibitedNo AI tools may be used to produce any part of the assessed work
Allowed with disclosureYou may use AI tools for stated purposes, and you must declare how you used them
Allowed for specific tasks onlyFor example, language editing or brainstorming is allowed, but generating text is not
Course-by-courseEach module or supervisor sets its own rule
Unclear or silentThe policy has not caught up, so you must ask

The safest approach is simple: find the rule, follow it and keep a record.

A sensible routine

  1. Read your university’s policy and your department’s thesis handbook.
  2. Ask your supervisor in writing if anything is unclear, and keep the reply.
  3. Use AI only for the purposes that are permitted.
  4. Disclose use in the way your institution requires. This may be a statement, a footnote or a form.
  5. Keep notes on which tool you used, when and for what.
  6. Check every fact, number and claim against a reliable source. You are responsible for the final work.
  7. Never cite a source you have not read. AI tools can invent references that look real, including plausible authors, journals and DOIs.
  8. Do not upload confidential or ethically restricted data (for example, identifiable participant data) to a public AI tool. Check your ethics approval and data protection rules.

An illustrative disclosure statement

If your university allows AI use with disclosure, a short statement might read as follows. Adapt it to your institution’s required wording.

‘I used an AI tool to suggest alternative wording for two paragraphs in Chapter 2 and to check grammar in Chapter 5. All ideas, analysis, sources and final wording are my own. I verified every reference against the original source.’

A clear, honest statement is far better than silence if your use is permitted.

Referencing AI tools

Some styles and universities give formats for citing AI tools, and some do not allow them to be cited as sources at all. If you need an example format, see our APA 7th edition referencing guide and our Harvard referencing guide for Nigerian students. Always check your department’s policy first.

If Your Own Writing Is Flagged

Being flagged when you wrote the work yourself is frightening. Here is what to do, step by step.

Step 1: Stay calm and do not delete anything

Do not panic, do not rewrite the flagged sections in a rush and do not delete drafts, notes or files. Your evidence is in them.

Step 2: Understand exactly what was flagged

Ask which passages were highlighted and what the percentage was. Check whether the result is an asterisk (below 20%), which Turnitin itself treats as low confidence, or a percentage of 20% or above.

Step 3: Collect your evidence

EvidenceWhy it helps
Dated drafts and version history (for example, Word or Google Docs history)Shows the work developing over time
Outlines, mind maps and plansShows your thinking before the text
Annotated PDFs and reading notesShows that you read and engaged with the sources
Source log or reference manager libraryShows how and when you found sources
Emails and feedback from your supervisorShows the project developing with oversight
Earlier writing samplesLets a reader compare your style
Data files, analysis outputs and field notesShows the original work behind the chapters

Step 4: Prepare to explain your process

Be ready to describe how you wrote the flagged sections: where you found the sources, how you took notes, how you structured the argument and what you changed after feedback. Be specific and truthful. Inconsistent explanations raise concerns more than honest, detailed ones.

Step 5: Be ready to discuss the content

Read the flagged passages again and make sure you can explain every claim, term and source in them. Knowing your work is your strongest defence. Our list of 40 dissertation defence questions can help you practise.

Step 6: Talk to your supervisor early

Ask for a meeting, bring your evidence and be open. A short, calm email can start the conversation.

Subject: Request to discuss my AI writing report

Dear Dr [Name],

I have seen that my submission for [assignment/chapter] has an AI writing result of [score/asterisk]. I wrote this work myself, and I would like to explain my process and show you my drafts, notes and sources. Could we arrange a meeting this week?

I am also happy to talk through any passage you would like me to explain.

Thank you, [Your name and student number]

Step 7: Know the process and your rights

Find your university’s academic integrity procedure and appeal route in the student handbook. Note the deadlines. If you feel you are not being heard, contact your students’ union, the postgraduate office or the ombudsperson, if your institution has one.

Step 8: Learn from it

Whatever the outcome, change your habits so that you can show your process in future. The next section explains how.

How to Protect Yourself: Build a Visible Writing Process

The best protection against both false accusations and real misconduct is a documented process. It costs little and pays off many times over.

  1. Write in a tool with version history, such as Word with autosave or Google Docs, and do not paste in finished text from elsewhere.
  2. Save dated drafts at the end of each work session or each chapter stage.
  3. Keep an outline or plan before you write each section.
  4. Annotate your sources and keep your notes, with author, year, page and a line on why the source matters.
  5. Use a reference manager such as Zotero or Mendeley, and check the entries by hand.
  6. Keep a short research diary with dates, decisions and questions for your supervisor.
  7. Share drafts with your supervisor on a regular schedule and keep the feedback.
  8. Keep your data files and analysis outputs safe and labelled.
  9. Record any permitted AI use in the same notes, with the tool, date and purpose.
  10. Write from understanding, not by copying and rewording. See how to paraphrase properly.

Writing Habits That Keep Your Own Voice Visible

The aim here is not to write ‘differently enough to fool a detector’, which is the wrong goal. The aim is to produce better academic writing, which happens to be writing that is clearly yours.

  • Be specific. Name your sample, your location, your dates, your instruments and your findings. Generic statements could be written by anyone, while specific ones could only be written by you.
  • Show your reasoning. Explain why you made a choice, not only what you chose.
  • Use your own examples, drawn from your own data, setting or experience.
  • Engage critically with sources. Compare authors, point out limits and say where you agree or disagree. See how to write a literature review.
  • Link chapters together. Refer back to your objectives and forward to your conclusions. See research objectives, questions and hypotheses.
  • Vary your sentences naturally without forcing awkward constructions.
  • Keep your meaning exact. Precise, accurate claims are better than smooth, vague ones.
  • Read your work aloud. Passages that sound unlike you often are.

Good writing of this kind also lowers your similarity percentage and improves your marks, which is a better reason to do it than avoiding a flag.

Humanisers, Spinners and ‘AI Bypass’ Tools: Why to Avoid Them

Search results are full of tools that promise to ‘make AI text undetectable’ or to ‘humanise’ it. Students should avoid them for several reasons.

  1. They can change your meaning. A claim that was accurate may become wrong or vague.
  2. They produce awkward or unnatural wording that your supervisor will notice.
  3. They do not fix invented references or false facts.
  4. They can still be detected, and vendors are actively working on detection of altered text.
  5. They are treated as deliberate evasion under many university policies, which can make a case worse.
  6. You cannot explain the result. If you are asked about a paragraph that a tool produced, you may not be able to defend it.
  7. Some of them store or reuse what you upload, which creates privacy and repository risks.
  8. Some are scams that take your money or your document.

If your work needs improvement, the honest route is to rewrite it from understanding, add your own analysis and get feedback from your supervisor or a reputable editing service that works within your university’s rules. See also How to Lower Your Plagiarism Percentage Ethically.

Students Who Write English as a Second Language

Many Nigerian students write academic English as a second or third language, or in a variety of English that differs from the one the detector was trained on. Some researchers and universities have raised concerns that AI detectors can be less reliable for writers working in a second language, and Turnitin has been asked about this publicly. There is no need for alarm, but there are sensible precautions.

  • Keep full drafts and notes, as described above.
  • Mention the concern to your supervisor early if you are worried, so that any flag can be put in context.
  • Do not ‘simplify’ or distort your natural writing to try to appear human. Clear, accurate writing is what matters.
  • If you use grammar or translation tools, check the policy and record what you used them for.
  • Ask a trusted reader to review your drafts for clarity.

A good supervisor understands that a student’s own voice includes the way they naturally structure sentences, and will judge the work in context.

AI and Your Dissertation: Chapter-by-Chapter Integrity

Different chapters carry different risks. This table summarises where to be careful.

ChapterWhere AI-related and integrity risks ariseSafer practice
Chapter 1: IntroductionGeneric background and a problem statement that could fit any studyWrite the problem around your own context and cite real evidence. See how to write a problem statement
Chapter 2: Literature reviewSmooth summaries of sources you have not read, and invented citationsRead each source, synthesise by theme and verify every reference
Chapter 3: MethodologyGeneric descriptions that do not match what you actually didDescribe your real sample, tools and procedures. See sampling techniques and sample size
Chapter 4: Results and discussionInterpretation that does not match your dataInterpret your own tables and figures and link them to the literature. See how to write Chapter 4
Chapter 5: Conclusion and recommendationsGeneric recommendations unrelated to your findingsBase each recommendation on a finding. See how to write Chapter 5
Abstract, references, appendicesUnverified references and unexplained instrumentsCheck every reference exists and cite every instrument. See APA vs MLA vs Chicago for style choices

Preparing for the Viva

Many universities use a viva or defence partly to confirm that the student understands and owns the work. If you can answer questions about your choices, sources and findings, the question of how the text was drafted rarely arises. Practise explaining each chapter in your own words, and prepare for questions about your methodology, limitations and contribution. See our guide to 40 dissertation defence questions and our dissertation writing and thesis writing pages.

Myths vs Reality

MythReality
‘Turnitin will tell my lecturer exactly which chatbot I used.’It estimates how much text resembles AI writing. It does not identify the tool or prove use
‘A low AI score proves my work is original.’No. The tool can miss AI text, and originality also depends on sources, citations and authorship
‘A high AI score proves I cheated.’No. Turnitin says the score should not be the only basis for action, and false positives are possible
‘The AI score and the similarity score are the same thing.’They are separate tools with separate reports
‘If I cite my sources, I cannot be flagged.’The AI indicator looks at writing style, not citations
‘Humaniser tools make AI text safe.’They can change meaning, can still be flagged and may count as deliberate evasion
‘Every university uses the AI indicator.’No. It depends on the institution’s licence and settings
‘Short or formal writing can never be flagged.’Formal, formulaic text can be flagged, and short texts may not be scored at all
‘Only students who cheat need to worry.’Honest students can also be flagged, which is why a documented process matters
‘I can check my work on any free AI detector and know my Turnitin result.’Different tools use different models, and only your institution’s report counts

Common Mistakes Students Make About AI Detection

  1. Assuming there is a safe percentage. There is no universal safe number.
  2. Not reading the policy. Many problems arise from not knowing what is allowed.
  3. Using AI for a task that the policy bans.
  4. Forgetting to disclose permitted use.
  5. Citing references that an AI tool suggested without checking them.
  6. Pasting AI output into a draft without rewriting or understanding it.
  7. Relying on humaniser or spinner tools.
  8. Deleting drafts and notes, which removes your best evidence.
  9. Uploading private data to public tools.
  10. Panicking and rewriting everything after a flag instead of gathering evidence and speaking to the supervisor.
  11. Treating unofficial AI checkers as authoritative, or uploading a whole dissertation to a site that stores it.
  12. Leaving everything to the last week, when stress leads to shortcuts.

AI Writing Readiness Checklist

Tick each item before you submit:

  • Do you know your university’s and your department’s AI policy?
  • Have you asked your supervisor about anything unclear, and kept the reply?
  • Is every use of AI within the permitted purposes and properly disclosed?
  • Have you written the text from your own understanding and in your own voice?
  • Have you checked every reference against the original source?
  • Have you verified every fact, figure and claim?
  • Do you have dated drafts, outlines, notes and a source log?
  • Can you explain and defend every section in a conversation or viva?
  • Have you avoided humaniser and spinner tools?
  • Have you read your similarity report as well? See What Is a Good Turnitin Similarity Score?
  • Have you kept confidential data out of public AI tools?
  • Do you know your university’s process if work is flagged, and the appeal route?

Frequently Asked Questions

Does Turnitin detect ChatGPT? Turnitin’s AI writing indicator is designed to flag text that resembles AI-generated writing, including output from popular chatbots, but it is an estimate and it is not perfect. It does not identify which tool was used.

Does Turnitin detect other chatbots and AI tools? The indicator looks at the style of the text, not at the brand of tool. How well it performs on a given tool can vary, and detection models are updated over time.

Is the AI score the same as the plagiarism score? No. The similarity score matches your text against other sources. The AI indicator estimates whether the writing resembles AI-generated text. They are separate reports.

What does the asterisk mean? It means that the tool found some AI-like text, but the result was above 0% and below 20%. Turnitin treats results at that level as low confidence, shows an asterisk instead of a number and does not highlight text.

What percentage of AI is acceptable? There is no universal figure. Policy is set by your university, and many treat the score only as a prompt for review. Ask your department what applies, and what rules cover AI use in general.

Can Turnitin wrongly flag my work? Yes. False positives are possible, especially at low percentages and for formal, formulaic or heavily edited writing. This is why Turnitin says a result should not be the only basis for a misconduct decision.

Can Turnitin miss AI-written text? Yes. AI text that has been rewritten by a person, mixed with human writing or altered by other tools can be missed, and results are less reliable for that kind of text.

Can I check my AI score before I submit? Only if your university provides it. Be cautious with unofficial checkers, because they use different models, their results do not predict your official result and some store or share your text.

Do grammar tools like Grammarly count as AI? Policies differ. Basic spelling and grammar correction is often allowed, while tools that rewrite whole sentences or generate text may fall under AI rules. Check your policy, and record what you used.

What if I used AI only for ideas or an outline? Check the policy. Some universities allow brainstorming with disclosure, and others do not. Either way, the final writing and analysis should be your own, and you should be able to explain every idea.

What if my university has no AI policy? Ask your supervisor, and ask for the answer in writing. Until you have clear permission, assume that submitting AI-generated text as your own is not safe.

What should I do if I am accused? Stay calm, gather your drafts, notes and sources, ask what was flagged, request a meeting and follow your university’s process. Do not delete anything. See the step-by-step section above.

Will my supervisor see my AI score? It depends on how your institution has set up the tool. Ask your department whether supervisors and students can see the result.

Does a high similarity score mean AI was used? No. A high similarity score means that text matches other sources, which can come from quotations, references or copied text. It says nothing about AI.

Can someone help me review and improve my own writing? Yes. Professional review and editing is legitimate when it follows your university’s rules and helps you understand and improve your own work. Our AI content detection and revision service helps you review and revise your own writing, citations and referencing, so that the final work reflects your own voice and meets your department’s standards.

Related Reading Across the Cluster and the Dissertation Series

If you are working on originality and integrity:

  • What Is a Good Turnitin Similarity Score?
  • How to Lower Your Plagiarism Percentage Ethically
  • How to Paraphrase Properly: Techniques and Examples
  • How to Write an Annotated Bibliography, Step by Step

If you are fixing citations:

  • APA 7th Edition Referencing Guide With Examples
  • Harvard Referencing: Complete Guide for Nigerian Students
  • APA vs MLA vs Chicago: Which Citation Style Should You Use?
  • OSCOLA Referencing for Law Students: Quick Guide
  • Vancouver Referencing for Nursing and Medical Students

If you are writing a dissertation:

  • How to Write a Literature Review
  • How to Write a Problem Statement
  • Research Objectives, Questions and Hypotheses
  • Sampling Techniques and Sample Size
  • How to Write Chapter 4: Results and Discussion
  • How to Write Chapter 5: Conclusion and Recommendations
  • 40 Dissertation Defence Questions

For Turnitin’s own explanation of its tools, see Turnitin’s guidance pages and Turnitin’s website. For free general guidance on paraphrasing, citing and academic integrity, Purdue OWL is a helpful reference.

Need Help Reviewing Your Draft?

If you are worried about how your work reads, an expert review can help you strengthen your own voice, fix weak paraphrasing and check your citations before your examiners see it. Writely.ng can review your draft, point out generic or unsupported passages, repair referencing and help you present your own work clearly, always within your university’s rules. Our related services include AI content detection and revision, Turnitin similarity reduction, APA formatting, Harvard referencing, dissertation writing and thesis writing.

Visit our Citation and Referencing Services page or contact Writely.ng for a free quote today, and submit work you can stand behind.


How to Lower Your Plagiarism Percentage Ethically
Previous Article
How to Paraphrase Properly: Techniques and Examples
Next Article

Abuja Office

16, Ademola Adetokunbo Crescent,
Wuse 2, Abuja Nigeria

Lagos Office

4th Floor, Suite 3AB, No 34/36 Ikorodu Road, Jibowu, Lagos

Port Harcourt Office

129-132 Old Michelin Compound Trans Amadi ind.Layout, Port harcourt Rivers

Ibadan Office

TDL House, 4 Ososami Rd, New GRA, Ibadan 200252, Oyo, Nigeria

“Exceptional service! Writely's content writing services in Nigeria are professional, reliable, and results-driven. Highly recommended!.”

Joaan Hensen

Request a free
consultancy

Writely offers top-notch content writing services in Nigeria, UK, US, UAE, and 103 countries; delivering expertly crafted content tailored to your unique needs. Our team specializes in creating high-quality, engaging, and SEO-optimized content to help businesses stand out. Trust Writely for professional content and writing services in Nigeria that drive results and elevate your brand presence.
Get In Touch
writely ng logo

Writely delivers exceptional content writing services in Nigeria, UK, USA, UAE, Canada and 105 countries

we pride ourselves on providing exceptional content writing services in Nigeria, Our professional content and writing services in Nigeria ensure your message resonates and delivers measurable impact.

Twitter Linkedin Instagram

Explore

Home
About Us
Our Services
Portfolio
Contact

Clients

Membership Login
Privacy Policy
Terms of Service
Table of Contents

Address

Email: [email protected]
Suite 3AB, 4th Floor,  No 34/36 Ikorodu Road, Jibowu, Lagos

Copyright 2025-31 by Writely Content Services All Right Reserved.