What AI is and what is not magic · Kode bersama Martin

AI stacks words. It is not a spell.

When you open a chatbot, it looks as if a clever friend is sitting there. It answers fast, politely and with certainty. That is a trap. It is not a friend. It is not a teacher. It is not magic. It is a program that stacks words that often fit together. It does it so well that it sounds like a person. That is exactly why you must know what it is not. AI does not know what is true. It cannot see your notebook. It does not know your teacher. It will not open the textbook on page 47. It guesses the next word. Sometimes it hits. Sometimes it invents a date, a name and a book that does not exist. In this course you do not learn code. You learn to ask, check and cite. When you can do that, AI is useful. Until then, it is a polite storyteller that sometimes lies without blushing.

Ema needs three sentences about photosynthesis for tomorrow. She types into the chat: “Write it nicely for me.” In five seconds she has a paragraph that sounds like a textbook. Adam looks over her shoulder and says: “That is magic.” Ema hands it in. The next day the teacher shows that the text has a scientist’s name that nobody in Year 6 is taught, and a year that is not in the book. Ema goes red. Adam whispers: “But it sounded clever.” That is the point. A clever-sounding sentence is not proof. It is a word puzzle. This time Ema opens the textbook, finds the chapter about leaves and writes three sentences herself. She leaves the chatbot for checking verbs — not for authorship.

💡

Do not say “AI knows”. Say “AI wrote”. Knowing lives in the textbook and in a person. A chatbot stacks sentences.

Guessing the next word, not a crystal ball

A model is a program trained on a lot of text. It learns which words often stand next to each other. When you write a start, it fills in a continuation that sounds natural. That is not searching Google. A search engine shows you links that exist on the web. A chatbot can show you a sentence that is nowhere — it just fits the previous words. That is why it sounds sure even when it is making things up. It is not evil. It has no conscience. It has nothing to be ashamed of. You are the referee. You open a book, a wiki, an atlas, or you ask a teacher. When you get this in the first lesson, you save yourself embarrassment at the board. AI is a tool like a calculator: it computes fast, but you must know if the result makes sense. A calculator does not know you meant to add apples, not cars. A chatbot does not know the teacher wants your words, not someone else’s paragraph.

Compare three things on one line. So you know what to reach for.

Tool	What it can do	What not to trust blindly
Textbook	Facts approved for school	It can be old, so check the date elsewhere too
Search engine	Links to pages that exist	The first link does not have to be true
Chatbot	A fast sentence, an outline, an example	A name, a date, a citation, a “study”

Example: you write “When did Charles IV die?” The textbook and the wiki say 1378. The chatbot usually hits that. You write “Who wrote the school article about our class in 2019?” The chatbot invents a name, because it cannot see that page. Same tool. Different certainty.

⚠️

Do not let certainty lull you. The more politely it sounds, the more easily you forget to check. A sure tone is not a mark.

Why this right now

You are in the course Ask with Martin, level Persiapan. 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 Inventor 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: AI stacks words — you watch the truth. When you can say it to a parent or a teacher, you have got the core. The rest is practice, not magic.

Without this sentence you will use the chatbot all course as a homework machine. Then a teacher, a parent or embarrassment arrives. Persiapan is there so you do not start with a bad habit. First you know what it is. Then you learn to ask. Only then you build a report, training and a code. When you say out loud today “this is not magic”, you have the base all the later lessons stand on.

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.

🏫 A report with citations

An outline, three sources that exist, and a sentence in your own words. AI must not be the only author.

No report yet. Just a sentence in the notebook: AI is not an author, it is a draft.

📝 Test training

Coach role: let it ask, do not let it say the answer. You count / answer yourself.

A test coach only makes sense when you know you must work out the answer yourself.

📅 Family planner

Tasks for the week (not passwords, not marks). A parent sees the plan, not your chat.

A planner is a list of tasks. It is not a fortune-teller that knows your week.

📜 An AI code for the class

Seven short rules on a poster. No horror — a how-to.

The first rule of the code: AI is not magic and not a teacher.

Workshop: one sentence, no spell

You open a chatbot with a parent. You are not handing in homework. You try whether you can say what it is, even when it writes it nicely.

Sit at the computer with a parent or a teacher. Open a chatbot you already have (ChatGPT, Gemini or Claude). If there is no account yet, stop — setting one up is the third lesson. Today you only read and try a question.
Type into the box exactly: “I am a school pupil. Explain in one sentence what you do, without the word magic and without foreign abbreviations.” Nothing else. No name, no school, no email.
Click send. Read the first sentence out loud to a parent. Ask: does it sound like a person, or like instructions? If it sounds like a person, write in your notebook: “It sounds like a person, but it is a program.”
Second question in the same chat: “Are you a teacher? Are you a search engine? Answer in two sentences.” Compare with the table in this lesson. If it says it is a teacher, you know it is not — a teacher gives marks and sees your notebook.
Tutup the chat. In the notebook write one sentence in your own words about what AI is. The word magic must not be there. When you can do that without looking, this lesson sits.

Practice: three nos

On paper three lines. On each write what AI is not. Then compare with a friend — not with the chatbot.

Line 1: “AI is not …, because …” Fill in teacher. The reason must be about a mark, a notebook or the class — not about it being stupid.
Line 2: the same about a search engine. Hint: links versus a made-up sentence. If you do not know, go back to the table above and rewrite it in your own words.
Line 3: the same about magic. Magic does not make a mistake you can find in a textbook. AI does. That is proof it is not a spell.

Now you 💪

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

You can say out loud that a chatbot stacks words and cannot see your notebook.
You know the difference between a search engine (links) and a chatbot (a sentence that may not exist).
In the notebook you have one sentence in your own words, without the word magic.

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.

✅

Check: You told a parent in one sentence what AI is not, and you have it written in the notebook.

What to take from this lesson

A chatbot sounds sure even when it does not know. A sure tone is not a mark of truth.
AI is not a teacher, a search engine or magic. It is a word stacker.
You are the referee: textbook, wiki, a person.
In this course you do not write code. You learn a question, a check and a citation.
First habit: do not say “AI knows”, say “AI wrote”.
The names in the examples (Ema, Adam) 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.