What AI is and what a chatbot can do

Why this matters for you

You may already hear people in your office talk about AI. Some say it will do everyone's work. Some say it is dangerous. Both are too simple. AI is a tool. Like a photocopier or a computer, it helps when you use it well. It causes trouble when you trust it too much.

This module explains AI in plain words. It does not ask you to use any tool yet. It only helps you understand what you are dealing with.

What AI means

Artificial intelligence or AI, is a name for computer programs that do tasks we usually think need a person. Examples are reading handwriting, understanding speech or writing a sentence.

AI is not a person. It does not think or feel. It does not know your office, your officer or your rules. It works by finding patterns in very large amounts of text and data.

What a chatbot is

A chatbot is an AI program you talk to by typing or sometimes by speaking. You write a question or an instruction. It writes an answer back. The message you type is called a prompt.

Most chatbots run on a large language model. This is a program that has read a huge amount of text. It has learned which words usually come next. So it is very good at writing smooth, confident sentences. But it is guessing the next words. It is not checking facts.

What a chatbot can do well

A chatbot can help you:

• write a first draft of a simple letter or reply
• turn rough notes into neat sentences
• make a long passage shorter and simpler
• change Roman Urdu into Urdu script
• give a rough translation between English and Urdu
• put a messy list into order
• explain a word or idea you do not understand

These are all tasks where you can read the result and check it yourself.

What a chatbot cannot do

A chatbot cannot:

• know the facts of your office. It does not know your file numbers, your rules or what your officer decided.
• always tell the truth. Sometimes it makes things up and sounds sure about them. This is called a hallucination. It might invent a rule, a date, a fee or a form number.
• know about recent events. Many chatbots learned from text up to a certain date. They may not know about new rules or changes.
• take responsibility. If a letter is wrong, the chatbot is not blamed. The person who sent it is.
• make decisions. It must never decide who gets a service, a payment or a document. People decide.

A simple way to think about it

Think of a chatbot as a quick new helper who joined the office yesterday. This helper writes fast and neatly. But the helper does not know your office. The helper never says "I do not know". Instead, the helper sometimes guesses.

You would let such a helper write a draft for you. You would never let them send it without reading it first. Treat a chatbot the same way.

Free tools and office rules

Many chatbots are free to use on a phone or computer. Free does not mean safe for office work. What you type may be stored by the company that runs the tool. The next module explains what you may and may not type.

Before you use any AI tool for work, ask your officer what your office allows. If there are no clear rules, ask before you start. For more background, see the guide on how chatbots work.

Try it

On your own phone or home computer, open any free chatbot. Ask it: "Explain in five simple sentences what a computer virus is." Read the answer. Then ask: "What is the fee for a new national identity card in Pakistan?" Do not trust the answer. Write down whether the chatbot said where the figure came from and whether it said it might be out of date. This shows you why facts must always be checked from an official source.

Check yourself

1. In one sentence, what is a chatbot?
2. What is a hallucination?
3. Who is responsible if a letter written with a chatbot has a mistake?

Answers: (1) A computer program you type to that writes answers back by guessing which words come next. (2) When a chatbot makes something up, such as a rule or a number and presents it as true. (3) The person who checked and sent it, not the chatbot.