Fundamentals and safe use
What an AI chatbot is, and what it gets wrong
Chatbots write likely-sounding text. That is useful for drafts and risky for facts.
It writes likely text, it does not look things up
A chatbot such as ChatGPT, Gemini or Claude was trained on a huge amount of text. When you ask something, it writes the answer that sounds most likely, one word after another.
That makes it very good at drafting, rewriting, summarising and explaining. It does not make it a database of facts.
It can be wrong and still sound sure
Chatbots sometimes state wrong numbers, wrong dates and wrong rules in a calm, confident voice. They can also invent sources, such as a law or a web page that does not exist. This is often called a hallucination.
Rule of thumb: the more a wrong answer would cost you, the more you must check it. Money, law, health, visas and deadlines always need the official source.
Try it
Catch it out
What is the current fee for a new Pakistani passport, and where do I pay it?
- Ask this to any chatbot you already use.
- Now open our passport fee guide and compare, line by line.
- Note anything the chatbot got wrong, left out, or gave without a date.
What to look for: Fees and rules change, and a chatbot may repeat an old figure. If its answer has no date or source, treat it as a lead to check, not as the answer.
Check yourself
A chatbot tells you the exact fee for a government service. What should you do before paying?
Show the answer
Check the fee on the government body's own page, or a dated guide that cites it. The chatbot's answer is a lead, not proof.
This course is free and always will be. Nobody pays to be included, ranked or recommended in it. Progress is saved only in this browser. Clearing site data resets it.