The prompt
Score 10 of 10, Excellent
Use it when You need to understand a study someone cited, what it actually found and how far you can rely on it, without a background in the field.
- {{paper_text}}
- {{decision_or_use}}
- {{claim_heard}}
- {{my_background}}
Full prompt and why it works
System prompt
You help people who are not researchers read academic papers. You explain what a study did and found in plain words, you keep what the paper says separate from your own general knowledge, and you never overstate what a single study shows.
User message
<paper> {{paper_text}} </paper> <context> What I will use it for: {{decision_or_use}} The claim I have heard made from it: {{claim_heard}} My background: {{my_background}} </context> Explain the paper above so that I can decide how much weight to give it in {{decision_or_use}}. I am not a researcher in this field, so define each technical term the first time you use it. Use these headings: 1. In one paragraph: the question the study asked, what the authors did and the main result, in under 100 words. 2. What they did: who or what was studied, how many, when and where, and the design (for example randomized trial, survey, interviews, or a model). Say what that design can and cannot show; a survey, for instance, can show a link but not a cause. 3. What they found: the main results with the numbers copied exactly, each with its section, table or page. 4. Limits: first those the authors state, then any others you notice, labeled "My reading" so I can tell them apart. 5. The claim I heard: does the paper support it, partly support it, or not address it? Quote the passage that decides it. 6. Questions to ask someone in the field: up to 3. Keep the whole explanation under 600 words. Use only the paper for what it says. Where you add general knowledge about the field, such as how this result compares with other research, mark it "(general knowledge, check)", and do not cite other studies from memory, because I cannot tell a real one from a misremembered one. <example> The form of one result only, with invented numbers: "Staff who used the checklist made fewer errors: 2.1% against 3.4% (Table 2), in one hospital over eight weeks." </example> A good explanation lets me repeat the finding accurately in one sentence, with the right amount of caution. If I pasted only the abstract, say so at the top, because the methods and limits may not be visible. If part of the paper is missing or a result is unclear, say what is missing rather than filling it in.
Variables
- {{paper_text}}
- The paper's full text, or at least the abstract, methods and results, with the title, authors and year.
- {{decision_or_use}}
- What you will use it for, for example "deciding whether to pilot a four-day week" or "a slide for the board".
- {{claim_heard}}
- What someone said the paper shows, in their words. Write "none" if you are reading it cold.
- {{my_background}}
- Your job and what you already know, for example "HR manager, no statistics training".
Why it works
- Background facts. Only the paper is used for what it says, and general knowledge is marked for checking, so you never repeat a finding the paper does not contain.
- What good looks like. Results come with exact numbers and their table or page, and the design is spelled out with what it can and cannot show, so the caution travels with the finding.
- What to make. A section on the claim you heard, with the deciding passage quoted, catches the common case where a paper is cited for more than it found.