AlvoreachStart free

Free resource · Guide

Research with AI without being misled

AI is great for getting your head around a topic, and terrible at giving you facts you can trust. Here is how to use it for the first without getting caught by the second — with prompts you can copy.

Free, and free of a catch. See the whole library.

Use AI to understand a topic, not to get facts

Think of AI as a well-read friend who is confident even when they are wrong. Ask them to explain a topic, lay out the main ideas, or tell you what to look into, and they are brilliant. Ask them for an exact number, a specific study, or a real quote, and they will happily make one up that sounds perfect. So use AI for the overview, and go find the actual facts yourself. It is a great place to start — just never the place you stop.

Start by asking AI to explain the topic

Instead of asking 'is this true', ask AI to explain the topic and point you where to look. Here is the prompt as a fill-in template, then a real example.

Copy and fill in the brackets
I want to understand [your topic]. Explain it to me in plain English: the main ideas, the key things to know, and the different opinions people have on it. Then tell me what a careful person would still need to check for themselves. Do not give me exact numbers or name specific studies — just help me understand the topic and know what to look into.
This plays to what AI is good at, and keeps it away from the facts it tends to invent.
Real example
I want to understand whether standing desks are actually good for you. Explain it to me in plain English: the main ideas, the key things to know, and the different opinions people have on it. Then tell me what a careful person would still need to check for themselves. Do not give me exact numbers or name specific studies — just help me understand the topic and know what to look into.
You get the shape of the debate and a to-do list of what to verify — not made-up statistics.

Make AI tell you what to double-check

After AI explains something, ask it which parts you should not take on trust. It is surprisingly good at pointing at its own weak spots.

Paste after any answer
Which parts of what you just told me should I not take at face value? List the specific things I should check for myself, and put the most important ones first.
Now you have a short list of exactly what to verify, instead of guessing.

Ask for search terms, never citations

If you ask AI for sources, it will often invent them — real-looking titles and links that lead nowhere. Ask for what to search instead. Real search terms you can actually check; fake citations you cannot.

Get pointers you can follow
Do not give me links or citations. Instead, tell me the best things to search for, and what kind of source would be trustworthy here — an official website, a well-known expert, the original report. I will look it up myself.
The one thing never to do: copy a link or a study name straight from AI into your work without opening it first.

How to check what you find

Treat anything AI tells you as a tip from a friend, not a fact. A tip is where you start looking — not something you repeat as true.

  1. 1Open it yourself. Never trust AI's summary of a page or study you have not read.
  2. 2Go to the original. Find the actual report, the official page, or the person who really said it — not someone describing it.
  3. 3Check it twice. If something matters, find it in a second, separate place. One source is a claim; two is a pattern.
  4. 4Mind the date. AI may not know about anything recent, even when it answers with total confidence.

Where AI most often gets it wrong

  1. 1Made-up sources: invented studies, authors, and links that look completely real.
  2. 2Wrong attributions: a real-sounding quote pinned to the wrong person or year.
  3. 3Nearly-right numbers: a figure that is close but not correct, because it blended a few similar ones together.
  4. 4Confident guessing: on anything niche or recent, a smooth answer that hides how little it actually knows.

The whole thing in four steps

  1. 1Understand the topic: ask AI to explain it and tell you what to look into.
  2. 2Get your checklist: ask which parts you should not just take its word for.
  3. 3Go and check: open the real sources yourself, and confirm anything that matters twice.
  4. 4Write from what you actually checked — and be honest, even with yourself, about what you did not.

More like this, as we write it.

The library is still filling. Start free and we will tell you when the next resource ships — one email, no spam.

Start free