How to Humanize an IB Extended Essay After ChatGPT

The IB Extended Essay is not a long English paper. Examiners score five criteria—focus and method, knowledge and understanding, critical thinking, presentation, engagement—against your subject guide and the research you actually did. ChatGPT extended essays hit word count, stack transitions, and often invent studies, statistics, and primary sources because the model completes the shape of research, not your reading log.

Your supervisor notices when criterion C analysis quotes a journal you never opened. Humanizing without fixing evidence produces polished criterion failures.

Related: argumentative essay claim + warrant, research abstract methods locked, literature review synthesis, ethical AI checklist.

What IB examiners and supervisors punish

Criterion	Weak AI draft	Strong reading-based draft
A: Focus and method	Broad topic, vague RQ	Narrow RQ scoped to available sources
B: Knowledge	Textbook summaries	Subject-specific concepts applied to your material
C: Critical thinking	"This shows" without analysis	Argument tied to evidence you can point to
D: Presentation	Template structure	Required sections, consistent citations
E: Engagement	Generic reflection	Specific decisions you made in the process

Supervisor comments usually arrive before the final draft. Treat them as a repair list for argument and sources—not as a prompt to regenerate the whole essay.

Research question + sources before any model

Before ChatGPT or Human Writes, write by hand:

1. Research question — one sentence, arguable, scoped ("How did X affect Y in Z period?" not "Importance of X")
2. Source list — every book, article, or data set you actually read or collected, with access date
3. Best evidence — three quotes, data points, or artifacts you highlighted—with page, table, or file reference
4. Supervisor note — the one comment that blocks progress ("narrow RQ," "more primary sources," "analysis not description")

If you cannot fill (3) from material on your desk, you do not have an EE yet. Re-read the subject guide assessment descriptors.

Do not paste paywalled PDFs into consumer chatbots if your school forbids it. Work from legal excerpts, notes, and data you are allowed to store.

Typical EE structure (adjust to subject guide)

Most subjects follow a variation of:

1. Title page / RQ / abstract (if required)
2. Introduction — context, RQ, why it matters in the subject
3. Body — evidence → analysis → link to RQ (not topic summaries)
4. Conclusion — answer the RQ; no new evidence
5. Bibliography / appendices — raw data, charts, interview transcripts

Section	Humanize?	Why
Bibliography / citations	No	Integrity
Data tables / graphs	No	Accuracy
Quotes from sources	No	Verbatim requirement
Analysis paragraphs	After evidence locked	AI skips criterion C
Reflection (criterion E)	Light, one pass	Must sound like your process

Paste analysis and evaluation paragraphs into Human Writes once. Restore every citation, footnote, and number exactly as in your source.

Before and after (criterion C analysis)

AI body paragraph:

Furthermore, economic theory suggests that inflation has many causes. This clearly demonstrates the importance of monetary policy. Studies show that central banks play a crucial role in modern economies.

After RQ, source, and supervisor comment addressed:

Fisher's equation (Fisher, 1930, 29) links nominal and real rates—the relationship my RQ asks about for Turkey 2018–2022. When policy rates lagged CPI in Q2 2021 (CBRT bulletin, Table 4), real rates turned negative; that pattern supports my argument that the bank traded inflation credibility for short-term growth, not that "policy is important" in general.

Second version has a subject concept, evidence from a source you can produce, and analysis that answers the RQ—not a summary of "economics."

Workflow with supervisor comments

1. Read the subject guide assessment descriptors for criteria A–E.
2. Submit an early draft your supervisor can mark—argument problems surface early.
3. Turn supervisor comments into a checklist: RQ scope, source type, analysis depth, citation format.
4. Fix checklist items from your reading and data—not from "rewrite my extended essay."
5. Build a one-page evidence map: claim chunk → quote/data → reference → ten-word link to RQ.
6. Draft body sections from the map.
7. Write introduction, conclusion, and criterion E reflection yourself.
8. One Human Writes pass on stiff analysis prose only.
9. Walk every citation against your bibliography and physical highlights.
10. Run formatting checks (word count, margins, required headings).
11. Disclose AI use if IB or your coordinator requires it.

For shorter in-subject essays using similar skills, see argumentative essay claim–warrant discipline.

What not to do

• Invent interviews, surveys, or lab results you did not run.
• Change a quote or p-value through paraphrase or humanizing.
• Ignore supervisor feedback and polish prose instead.
• Let the model add sources not on your bibliography.
• Humanize five times until your RQ drifts from the subject guide.
• Treat a humanizer as a substitute for reading primary material.

Detector scores are a review signal at some schools, not an IB grade. Missing criterion C fails faster than a percentage. See best practices for humanizing.

Bottom line

Extended essays win on a scoped research question, sources you can defend, criterion C analysis, and supervisor feedback addressed. Human Writes is the voice pass after your evidence map is true—not invented research dressed up for examiners.

Paste stiff analysis paragraphs on Human Writes when every citation points to a source your supervisor can ask about in a viva.