EXPERIENCE AN AI TUTOR LESSON

Can you spot an
AI hallucination?

Learn the concept, practise with an exercise, complete your own task and review your work. Follow the full learning sequence below.

← Back to the product

Interactive demo · Fictional learning material. The exercise uses pre-written feedback; the written task has a model answer and guided self-review. Your work stays in this page and resets when you leave.

STEP 1 OF 4 · LESSON
01 / LESSON

First, understand the concept.

An AI hallucination is an answer that sounds plausible but contains false information. It might include an incorrect number, an invented quotation or a reference that does not exist. A confident tone does not make a claim reliable.

What you will learn

Check an AI answer against a source, distinguish errors from missing evidence and write a correction that keeps uncertainty visible.

Three ways to read a claim

Supported

The source confirms it.

Contradicted

The source says otherwise.

Not stated

We need more evidence.

A QUICK WORKED EXAMPLE

Reference: “The meeting begins at 10:00.”
AI answer: “The meeting begins at 09:00 and lunch is included.”

The start time is contradicted. Lunch is not stated—the reference alone cannot confirm or disprove it.

Your habit: separate the answer into claims, look for evidence and make any unresolved questions explicit.

Further reading: Why language models hallucinate ↗