Chatbot, model, prompt

Three words so you do not get lost in names

On the box they write ChatGPT, Gemini, Claude, Copilot. At school someone says model, someone bot, someone artificial intelligence. When you mix those names, you do not know what you control and what is hidden in the background. A chatbot is the window you type into. A model is the program at the back that stacks the answer. A prompt is what you sent — a question, context, a rule. A bad prompt is not “bad AI”. It is an unclear order. When you say “something nice” in a cake shop, you get a random cake. When you say “a small chocolate one for six people with no nuts”, you get what can be baked. It is the same here. These three words stay with you the whole course. Without them you will moan about the brand, instead of fixing the question.

Adam opens Gemini, Bára ChatGPT. Both write “help me with history”. Adam gets a long lecture about the Hussites. Bára gets a quiz. They argue which chatbot is better. A parent settles it: “Did you write the same thing? And did you say your year?” They did not. Adam is in Year 7, Bára in Year 5. The same three words, a different house. A chatbot is the door. A model is the cook in the kitchen. A prompt is the order. When the order is foggy, the cook guesses. Bára writes on a slip: chatbot = window, model = program, prompt = my sentence. Adam adds: “When we argue who is better, first we compare the prompt.”

Window, cook, order

A chatbot is a page or an app. It has a box, a send button, a chat history. You can see it. The model is hidden. It may have a number or a nickname. You usually do not pick it — the service picks it. The important bit: the model does not know today’s homework until you write it. It has no access to the school Wi-Fi or the school portal. A prompt is the whole bundle you send: role, year, subject, what you already know, what you do not want. One word “history” is a weak prompt. The sentence “I am in Year 7, we are learning the Hussites, I want three revision questions, not a whole lecture” is a strong prompt. The more precise the order, the fewer hallucinations from guessing. The whole Curious and Little bird course is really prompt training. Today it is enough to be able to point at those three words on the screen.

Number it in the margin of the notebook. When someone asks, you show three lines.

1. Chatbot — what you click: the box, send, a new chat. Example: the ChatGPT window in the browser.
2. Model — the program at the back. You do not see it, like a cook behind a door. It changes when the company updates.
3. Prompt — your message. Even “hello” is a prompt. A bad prompt: “write it”. Better: “I am in Year 6, topic the water cycle, I want a comparison, not a definition from an encyclopaedia”.

Try with Bára the same prompt in two brands. If it comes out similar, it is not a miracle — it is the same game. If it comes out different, it still holds: first compare the brief, not the logo.

Why this right now

You are in the course Ask with Martin, level 준비. You work in a chatbot (ChatGPT, Gemini or Claude) — same game, different brand. No code. A parent or the school creates the account. A child’s email does not belong here.

This is not the 인벤터 level. There you build a chatbot on your own website. Here you are a user: a question, a check, a citation.

You will know this lesson by one sentence: A chatbot is the window, a model the cook, a prompt the order. When you can say it to a parent or a teacher, you have got the core. The rest is practice, not magic.

Without a glossary you will hear prompt in later lessons and not know it is simply what you type. A teacher will ask how you did it. The answer “in AI” is weak. The answer “I gave a prompt with my year and a ban on finished homework” is honest. Three words are the ticket into Curious.

Four things you will build in this course

The course has four small projects. You do not have to do them all at once. Today you add only what this lesson can do.

Workshop: show three things on the screen

You will not write a long task. You find on the screen the window, the box and one sentence you send on purpose twice — once foggy, once precise.

1. Open a chatbot with a parent. Point with a finger at the box where you type. Say out loud: this is the place for a prompt. Show the send button. Say: this sends the prompt to the model. Show the name at the top (ChatGPT, Gemini, Claude). Say: this is the chatbot, the brand of the window.

2. Write a weak prompt: “help with science.” 보내기 it. Read the first five lines. Are they general? Do they match your year? Usually not. That is the cook guessing without an order.

3. In the same chat write a strong prompt: “I am in Year 6. We are learning the water cycle. I want one comparison to water in a kettle, not a definition. No finished school homework.” 보내기 it.

4. Compare both answers. Which is shorter and to the point? In the notebook write: weak prompt = … / strong prompt = … In your own words, two commas are enough.

5. 닫기 the chat. Tell a parent the three words and what they mean. If they mix them up, correct them kindly. When they can explain it too, you can.

Practice: fix the order

Three foggy sentences. For each write a stronger prompt. Do not send them, paper is enough.

1. “Write it for me from maths.” Fix: year, topic (say fractions), what you want (one sample question, not homework).

2. “Let’s do a report.” Fix: subject, topic in one sentence, output (an outline of five points), a ban on finished paragraphs.

3. “Translate this.” Fix: from which language, into which, and that after the translation you must say the meaning in your own words.

Now you 💪

Go through three checks. At each one tick in your head: can I see it, or not?

1. You can point on the screen where the prompt is, where the chatbot is, and that you cannot see the model.
2. You have in the notebook an example of a weak and a strong prompt from the workshop.
3. You do not mix up the brand (ChatGPT) with what you type.

If something is missing, go back to the steps above. This is not a test for marks. It is a safety net, so you go on sure, not confused.

What to take from this lesson

• A chatbot is a window. A model is a program. A prompt is your order.
• A foggy prompt forces the model to guess. Guessing breeds hallucinations.
• Before you change the brand, change the prompt.
• A prompt belongs to privacy — do not share it if there was something personal in it.
• The whole course is largely training in how to ask.
• The names in the examples (Adam, Bára) are made up. Do not put your number, email or photo here.
• Next lesson: When you can do this, go to the next lesson in the same level.

© 2026 Ing. Martin Polak / AlgoRhino · 콘텐츠 이용 약관