Lesson 1 of 7 · AI for Everyone: Use AI Confidently at Work
What AI Chatbots Really Are (and What They Are Not)
A plain-language explanation of how AI chatbots work, what they do well, and where they go wrong.
Before you use any tool well, you need a rough idea of how it works. You do not need the maths. You only need the right picture in your head, because that picture decides how much you trust the answers.
What an AI chatbot is
ChatGPT, Gemini, Claude and Microsoft Copilot are all built on something called a large language model. It is a computer program that has read an enormous amount of text: books, websites, articles, code and more.
From all that reading it learned the patterns of language. When you type a question, it writes a reply by working out, word by word, what a good answer is most likely to look like.
That is why it can write a polite email, explain a GST concept in simple words or turn your notes into a neat table. It has seen millions of examples of each.
Think of it like this: an AI chatbot is a very fast, very well-read assistant who has never worked at your company and sometimes speaks confidently about things it does not actually know.
What AI is good at
| Good at | Examples |
|---|---|
| Writing and rewriting | Emails, letters, social posts, making text shorter or more polite |
| Explaining | "Explain this Excel formula", "What does this clause mean in simple words?" |
| Summarising | Long documents, email threads, meeting notes |
| Organising | Turning messy notes into tables, checklists or step-by-step plans |
| Brainstorming | Ideas for a campaign, names for a product, questions for an interview |
| Translating | English to Hindi and back, or into simpler English |
What AI is not
It is not a search engine. Unless the assistant is searching the web for you (some can), it answers from what it learned during training. That knowledge has a cut-off date, so recent news, new tax rates or this year's deadlines may be missing or out of date.
It is not a calculator or a database of facts. It can get numbers wrong, mix up names and even invent sources that do not exist. Lesson 5 shows you how to catch this.
It does not know your business. It has never seen your customers, your files or your company's rules. It only knows what you tell it in the chat.
It is not responsible for the result. You are. If an AI-written email goes out with a mistake, it is still your email.
Why the answers change
Ask the same question twice and you may get two different answers. That is normal: the model is generating text, not looking up a fixed reply. If an answer is not right, you can simply ask again, or better, ask more clearly. That is what lesson 3 is about.
Practice task
Open any AI assistant you can access today (if you do not have one yet, do this after lesson 2) and type:
I work as a [your job title]. Explain in 5 short bullet points what you can help me
with in my daily work, and 3 things I should NOT rely on you for.
Read the answer. Does the list of "not rely on" match what you learned in this lesson?
Key takeaways
- AI chatbots write answers by predicting what a good reply looks like, based on huge amounts of text.
- They are excellent at writing, explaining, summarising and organising.
- They can be confidently wrong, and their knowledge is not always up to date.
- They only know what you tell them, and you stay responsible for what you use.
Finished this lesson?Mark it complete to track your progress on this device.