20 research prompts that separate known from guessed

Models answer research questions with the same confidence whether they know the answer or not. These prompts make the model sort claims by evidence, say what would change its mind, and flag what you need to check yourself.

Swap anything in [brackets] for your question, sources or draft.

01. Map what is known before you search

Starting research on anything new.

I am researching: [question].

Before I search, give me:
1. What is well established, with the kind of source that establishes it.
2. What is debated, and the main positions.
3. What is unknown or too recent to be settled.
4. The search terms and sources I should start with.

Mark anything you are unsure about as [UNSURE]. Do not invent citations.

02. Sort an answer's claims by how well they are supported

You got a confident answer and need to know what to trust.

Here is an answer I got about [topic]:

[paste the answer]

List every factual claim in it. For each, rate the support: well established, plausible but needs a source, or likely wrong. Say what kind of source would confirm it and what I should search to check.

Do not add new claims.

03. Summarize a source without losing the caveats

A paper, report or long article you need the point of.

Summarize this source for [who and why they need it].

Include: the main finding in one sentence, how they got it (method, sample, data), the numbers that matter, and every limitation the authors mention. Then add limitations they did not mention but a critic would.

Quote the source for any number. Under 300 words.

[paste the source]

Unlock the other 17 prompts for free

Sign up for the Whizi newsletter to read the rest of this pack.

  1. Compare sources that disagree
  2. Plan a literature review
  3. Write an interview guide that avoids leading questions
  4. Find the patterns across interview notes
  5. Research a competitor from what they publish
  6. Build the strongest case against your view
  7. Understand a research method before you trust it
  8. Check a statistic before you repeat it
  9. Turn a vague request into a research brief
  10. Write an annotated bibliography entry
  11. Build a timeline from scattered sources
  12. Find the gaps in your research draft
  13. Prepare questions for an expert
  14. Estimate a market size you can defend
  15. Turn research into a summary for non experts
  16. Judge how reliable a source is
  17. Have a second model fact check the first one's research