Home · Personal · Learning & Study · Essays & study skills

YES

As of 13 August 2026, AI can extract themes from your research articles.

Most people should do this themselves; the prompt is on this page.

Can you do it?

5 minutesto a draft.

30 minutesto something you’d act on.

Cost, all in£0

Skill neededchat-fluent

Who has to check ityou

What the alternative costsA purpose-built PDF tool such as Humata can answer questions about long PDFs and provide cited answers; no comparable human-service price is supplied here.

If this goes wrong: you accept a weak or invented theme, but you can usually find and replace it by checking the cited passages against the articles.

What to actually do

  1. Do it yourself

    The route this page recommends

    A chat interface, chat-fluent skill, and roughly 30 minutes until you can act on the result.

    How to actually do it

    1. Open the articles and gather the full text or PDFs, keeping each article's title, author and page numbering visible.
    2. Write down the research question, assignment purpose and any required theoretical or coding framework.
    3. Upload the articles or paste them into a chatbot, then paste the copyable prompt and replace the bracketed slots with your question and materials.
    4. Ask the model to process the articles together and produce theme definitions, supporting passages, differences between articles and evidence gaps.
    5. Open each cited passage in the original article and mark whether it supports the proposed theme, contradicts it or has been overstated.
    6. Rename, merge or split themes where your checking shows that the definitions do not fit, then use the checked themes and quotations in your notes or essay.

    Prompt

    I am extracting themes from research articles for [essay, dissertation or study purpose]. Using only the articles and notes I provide, identify recurring ideas, concepts, experiences or arguments that are relevant to this research question: [research question].

For each proposed theme, provide:
1. A short, precise theme name.
2. A definition stating what belongs in the theme and what does not.
3. The main supporting passages, with article title and page number or section heading where available.
4. Which articles support the theme and whether they disagree or use different meanings.
5. Any useful subthemes.
6. Evidence gaps and passages that do not fit.

Keep explicit findings separate from your own interpretation. Do not invent quotations, page numbers, findings or connections. Do not treat a topic mentioned once as a theme unless you explain why it matters. Start with a table, then give a short explanation of how the themes answer the research question. Finish with a list of claims I must check in the original articles. Here are the articles and notes:
[PASTE OR UPLOAD ARTICLES AND NOTES]

    Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.

  2. Use a tool built for this

    Second choice

  3. Hand it to a person

    The distant third

    A person who owns the outcome does this end to end, worth it when the failure is dear.

What it gets wrong

  • AI cannot decide the final boundaries between overlapping themes without your research question and interpretation.
  • It can flatten differences between studies by grouping similar words that have different meanings in context.
  • It cannot replace close reading of passages where irony, disciplinary language or an author's qualification changes the meaning.
  • It may miss relevant themes in long or poorly formatted articles, even when it gives a tidy list with citations.

Even on a YES, the friction has a name: judgement under ambiguity and context depth.

How we scored this

Five axes, each scored nought to two by hand: ten means AI carries the task cleanly, and the thresholds that turn a total into YES, PARTLY or NO are published in the methodology. Each axis name links to its definition.

AxisScore (0–2)
Output2
Inputs2
Verification1
Liability2
Effort delta2
Total9 / 10

The methodology and its thresholds are published in full.

FAQ

Can AI find themes in research papers?
Yes, AI can produce a useful first-pass set of themes from uploaded papers and point to supporting passages. You still need to check the passages and decide whether each theme answers your research question.
How do I use AI to code research articles?
Give it the articles, your research question and clear instructions to label passages, define each code and show the evidence. Treat its codes as provisional, then compare them with the original text before turning them into themes.
Can ChatGPT analyse journal articles?
Yes, it can summarise and compare articles and propose recurring concepts or arguments. It may miss context or overstate a connection, so ask for quotations and page references and check them yourself.
Is AI reliable for thematic analysis?
It is reliable enough for organising material and suggesting patterns, but not as the sole basis for a defensible thematic analysis. You must make the interpretive decisions, check the evidence and follow your course or research method.

Nearby answers

All Essays & study skills rulings

Assessed by gpt-5.6-luna (gpt-5.6-luna) on 2026-08-13, second-checked by an independent model. Wrong somewhere? Email [email protected] and it gets re-checked.

The newsletter

AI news, new answers and product picks, straight to your inbox.