VERIFY / COMPARE

Verify Before You Trust

Don’t hand your brain to the first polished answer. Compare outputs, pressure-test claims, and verify facts when accuracy actually matters.

WHEN TO VERIFY

Verify facts, names, numbers, quotes, dates, and anything that sounds highly specific.

Slow down when the answer feels too smooth, too certain, or oddly convenient.

For school, money, health, legal, or major decisions, verification is not optional.

THE BASIC RULE

Compare when quality matters.

Verify when facts matter.

Use your own files first when the truth should come from your notes, docs, or sources.

RED FLAGS

When you should stop and check

These are the moments where confident nonsense shows up most often.

Research claims

If it names a study, source, quote, statistic, date, or historical claim, verify it.

Math and numbers

AI can sound sharp and still blow the calculation. Re-check the math and the logic.

High-confidence weirdness

If it sounds a little too perfect, too absolute, or too easy, treat that as a warning sign.

WORKFLOW

Compare two answers the smart way

Do not just pick the one that sounds better. Force the models to reveal what is missing.

Workflow

  • Ask the same question in two AIs.
  • Paste both answers side by side.
  • Ask which one is clearer, better sourced, and more complete.
  • Keep the best parts, not just the first answer.

Comparison prompt

Compare Answer A and Answer B. Tell me: 1) which is more accurate 2) which is clearer 3) what each one is missing 4) which answer I should trust more and why If neither is strong, say that clearly.
TOOLS

Use the right verification tool

Different tools are good at different kinds of checking.

Perplexity

Fast source hunt. Good when you need links and citations quickly.

Google Scholar

Better for academic papers and more serious research trails.

NotebookLM

Best when the truth should come from your own files and notes.

WolframAlpha

Good for math, calculations, and technical checks.

PROMPTS

Quick prompts to pressure-test an answer

Use these when you want the model to show its weak spots instead of hiding them.

Find the weak spots

What in this answer is known, and what is only being guessed? Show me what needs verification.

Source check

Give me sources for the main claims in this answer. If you cannot confirm a claim, say cannot confirm.

Bias / one-sidedness

Show me where this answer may be one-sided, overconfident, or missing another perspective.

Math / logic check

Check this answer for math errors, skipped steps, and logic gaps. Flag anything that does not actually follow.
BOTTOM LINE

Best rule

Compare when it matters. Verify when facts matter. Use your own files when they are the real source of truth.