📊 We use cookies & analytics to improve your experience. Learn more
Latest

Claude Mastery — From First Prompt to Advanced Workflows Module 4 · Lesson 14 of 15

0% Exit to course

Module 4 · Career, projects and honesty Lesson 14 of 15

Limits, hallucinations and how to check

Learning objective

These tools predict plausible text. It is careful, which is not the same as correct — a wrong number inside a careful paragraph is harder to spot, not easier That is a mechanism, not a moral failing — and the habit it demands is the habit of asking "where did this come from?" out loud, every time, before you use an answer.

The concept

A model does not look things up unless it has been given a way to look things up, and even then it can attach a real number to the wrong claim. It is careful, which is not the same as correct — a wrong number inside a careful paragraph is harder to spot, not easier

So the rule is: numbers, dates, names, citations and legal or medical claims get checked in a primary source before they leave your hands. Everything else — drafts, structure, explanations, practice — is fair game.

Three questions to ask before you trust an answer. Where did this come from? What would make it false? Who is the original source, and can I open it? If the third one has no answer, you have writing material, not facts.

On a long document, the fastest check is to ask for the page number with every claim and open two of them at random.

Worked example

Ask it to total a column in an attached spreadsheet and check the total yourself. Any model that is reading text can misread a column.

Your practice task

Ask for one statistic with its source, then open the source. Sometimes it exists. Sometimes the citation is invented, and the number is close enough to a real one to be dangerous. Either way you will remember the exercise.

Do this before moving on. Reading a solution teaches the shape of it; typing it teaches you.

Key takeaway

A model does not look things up unless it has been given a way to look things up, and even then it can attach a real number to the wrong claim.

Create free account or sign in

Almost there

15 lessons left of 15

Finish every lesson to unlock the certificate. Each one ends in a practice task — do them and the final project brief writes itself.