Artificial intelligence
How to check an AI answer before relying on it
A simple source, date and common-sense check before using AI-generated information.
What you will learn
You'll know a simple way to check source, date, context and risk before using an AI answer.
Before you continue
Do not rely on AI alone for high-stakes decisions or instructions that could harm someone if wrong.
Stop and use an official source or qualified professional if the answer affects money, health, safety, legal rights, employment, customers or another person.
Decide how much it matters
Not every AI answer needs the same level of checking. A dinner idea is low risk. Advice about money, health, legal rights, safety or customers is not.
The more the answer matters, the more carefully you should check it.
Look for the source
Ask yourself where the answer came from. If the AI gives a source, open it yourself and read the relevant part.
Do not assume a source exists or says what the AI claims.
Check the date
Some answers go wrong because information changed. Look for a review date, publication date or current official guidance.
This matters for prices, rules, software settings, security advice and local services.
Compare with an official place
For important information, compare the answer with an official website, provider support page, government agency or qualified professional.
If the official source says something different, use the official source.
Watch for invented detail
Be cautious when an answer gives exact names, quotes, links, statistics or steps that you cannot verify.
Smooth wording is not the same as accuracy.
Keep a person responsible
AI can help you organise a question or draft a first version. A person still needs to decide whether the answer is suitable and safe to use.
What to expect
You can decide whether an AI answer needs no check, a quick check or a reliable outside source.
Sources and further reading
- Why language models hallucinate · OpenAI
Supports the explanation that AI tools can produce plausible but incorrect claims.
- Artificial intelligence · eSafety Commissioner
Supports checking AI-generated material and understanding AI limitations.