Key Takeaways
- The discussion is the most intellectually demanding section of a research paper. LLMs can help structure it, improve flow, and check hedging language, but the interpretation must be yours.
- LLMs do not know your field deeply enough to compare your findings to the literature accurately. Always verify any comparison an LLM makes against the actual sources.
- The best use of LLMs for the discussion is structural: organising your interpretation, writing limitations paragraphs, drafting implications, and checking that every research question is addressed.
- These 10 prompts cover structuring the discussion, writing about limitations, comparing to prior work, drafting implications, and ensuring completeness.
- For a comprehensive guide to using LLMs across all stages of research, see our LLM Prompting Guide for Researchers.
The discussion section is where your paper shows its intellectual depth. You interpret your results, connect them to existing literature, acknowledge limitations, and explain what your findings mean for theory and practice. It is the section reviewers scrutinise most closely, and the one where LLMs must be used most carefully.
An LLM does not understand your research context. It cannot judge whether your results genuinely support a comparison to a previous study. It cannot evaluate whether your limitations are the ones a reviewer would actually care about. What it can do is help you structure your discussion, draft limitations paragraphs, check hedging language, and ensure you have addressed all your research questions.
For broader guidance, see our LLM Prompting Guide for Researchers.
The 10 Prompts
Prompt 1: Structure the Discussion
My paper has [number] research questions and the following key findings:
[list findings]
Suggest a logical structure for the discussion section. It should: (1) open with a summary of key findings, (2) discuss each finding in relation to a research question, (3) include a limitations subsection, (4) include implications for theory and practice, and (5) suggest future research. Provide subheadings and a one-sentence description for each subsection.
Prompt 2: Draft a Limitations Paragraph
My study has the following limitations:
– [limitation 1, e.g., “small sample size of 45 participants”]
– [limitation 2, e.g., “cross-sectional design, cannot establish causality”]
– [limitation 3, e.g., “self-reported data, subject to social desirability bias”]
Write a limitations paragraph that acknowledges each limitation honestly, explains how it may have affected the results, and uses appropriate hedging language. Do not dismiss the limitations. Treat them as genuine constraints on the interpretation of findings.
Prompt 3: Compare a Finding to Prior Literature
My finding: [describe your finding with data]
This finding is [consistent with / contradicts] the following previous studies:
– [Author, year: their finding]
– [Author, year: their finding]
Write a paragraph that: (1) states my finding, (2) compares it to these studies, (3) suggests possible explanations for any agreement or disagreement, and (4) discusses what this means for understanding [the broader topic]. Use hedging language where appropriate. I will verify the comparison for accuracy.
Prompt 4: Write Implications for Practice
Based on the following findings from my study:
[list key findings]
Write a paragraph discussing the practical implications. Who benefits from these findings? How could practitioners, policymakers, or educators apply them? Be specific rather than generic. Avoid “more research is needed” as an implication. Focus on actionable applications.
Prompt 5: Suggest Future Research Directions
My study examined [topic] using [methodology] with [population/context]. The main limitations were [list limitations]. Based on these limitations and my findings, suggest 3-4 specific future research directions. Each suggestion should address a limitation or extend the study in a meaningful way. Be specific: name the methodology, population, or variable that future studies should target.
Prompt 6: Check Hedging Language
Review the following discussion section and flag any claims that are too strong for the evidence presented. Look for language like “proves,” “clearly demonstrates,” “definitively shows,” or “certainly.” For each flagged instance, suggest appropriate hedging language (e.g., “suggests,” “may indicate,” “is consistent with,” “appears to support”).
[paste your discussion section]
Prompt 7: Write the Opening Summary
The discussion section should open with a brief summary of the key findings. My main findings are:
1. [finding 1]
2. [finding 2]
3. [finding 3]
Write a concise opening paragraph (3-4 sentences) that summarises these findings without repeating statistical details. The tone should transition from reporting (results section) to interpreting (discussion section).
Prompt 8: Explain an Unexpected Finding
My study found [unexpected result], which contradicts the hypothesis that [state hypothesis]. Previous literature suggested [what was expected and why]. Write a paragraph that: (1) acknowledges the unexpected nature of the finding, (2) proposes 2-3 possible explanations (methodological, contextual, theoretical), and (3) suggests what this means for the field. Use hedging language throughout.
Prompt 9: Check Research Question Coverage
My research questions are:
1. [RQ1]
2. [RQ2]
3. [RQ3]
Review the following discussion section and check whether each research question is explicitly addressed. For any question not addressed, tell me where in the discussion it should appear and what content is needed.
[paste your discussion section]
Prompt 10: Write the Theoretical Contribution
My study contributes to [theoretical framework/field] by showing that [key finding]. Previous understanding was [what was previously known]. My finding extends/challenges/refines this understanding by [explain how].
Write a paragraph that articulates this theoretical contribution in academic prose. Position the finding within the existing theoretical landscape. Use hedging language. Do not overclaim.
The Discussion Requires Your Intellectual Input
Of all the sections in a research paper, the discussion is the one where AI assistance is most risky. The interpretation of your results requires deep knowledge of your field, your data, and the specific context of your study. An LLM can help you structure and articulate your thoughts, but the thoughts themselves must be yours. Use these prompts as scaffolding, not as a substitute for critical thinking.
Need your discussion section reviewed before submission? PM Proofreading’s journal editing service ensures your discussion is clear, well-argued, and appropriately hedged. Upload your manuscript and get a quote.
Further Reading
- LLM Prompting Guide for Researchers: ChatGPT, Claude, and Gemini
- How to Structure and Organize a Research Paper
- How to Write an Effective Introduction Section
- AI vs Human Proofreaders and Editors
Frequently Asked Questions
Can I use AI to write my discussion section?
You can use AI to help structure, draft, and refine the discussion, but the interpretation of your results must be your own. An LLM does not understand your research context deeply enough to draw valid conclusions. Use these prompts as scaffolding for your thinking, not as a replacement for it. Always verify any comparison to previous literature against the actual sources.
How long should the discussion section be?
For a typical 5,000-8,000 word journal paper, the discussion is usually 800 to 1,500 words. It should be proportional to the complexity of your findings. A study with three research questions and multiple findings needs a longer discussion than a study with one focused question. Do not pad the discussion with generic statements.
Should I discuss every result in the discussion section?
Focus on the results that are most relevant to your research questions and most interesting to the field. Minor or expected findings can be mentioned briefly. Unexpected or contradictory findings deserve more space because they require explanation. If a result does not add to the narrative, it may not need discussion beyond what the results section already covers.
How do I discuss limitations without undermining my study?
Present limitations honestly but constructively. Acknowledge each limitation, explain how it may have affected results, and where possible, note how you mitigated it. Frame limitations as opportunities for future research rather than fatal flaws. Reviewers respect transparency. They become suspicious when a paper claims to have no limitations.
