Beginner · Level 1: fundamentals
What generative AI is (and what it is not)
Understand how a chatbot “thinks”, why it sounds sure even when it is wrong, and what you can use it for from day one.
What you will learn
- Explain in your own words what generative AI does
- Know why a chatbot can confidently write false things
- Choose which tasks are worth giving to AI and which are not
When you write to a chatbot and get an answer that sounds like a person wrote it, it is easy to believe that something on the other side “knows”. It does not, and that difference changes how you should use it. In this lesson we look, simply, at what actually happens.
What “generative” means
Generative AI is a program that creates new content: text, images, sound, code. It does not look up a ready-made answer in a database. It builds the answer piece by piece, based on patterns learned from huge amounts of examples. A chatbot such as ChatGPT, Claude or Gemini is, at its core, a language model: at each step it picks the word (or word fragment) that best fits everything written so far in the conversation.
Think of the autocomplete on your phone, but far more powerful. Phone autocomplete guesses the next word. A language model guesses the next word tens of thousands of times in a row, so whole coherent paragraphs come out.
Why it sounds sure even when it is wrong
The model is built to produce plausible text, not verified text. If you ask something it does not know, it often does not say “I don’t know”; it produces an answer that looks good: names, dates, book titles or links that seem real and do not exist. Specialists call this a “hallucination”. A firm tone is not proof of accuracy; it is simply the style it learned to write in.
A chatbot does not “search the internet” on its own. Some services have a search tool that they switch on only sometimes, and then they show their sources. If you see no sources, assume the answer comes only from what it learned, which may be outdated or incomplete.
What it does well
- Rewrites and summarizes: a long text becomes ten lines, a harsh message becomes polite.
- Explains complicated things at your level, and if you did not understand, you can start again with another example.
- Gives ideas and drafts: titles, the structure of an email, versions of a presentation.
- Helps you learn: asks you questions, corrects a sentence, explains a mistake.
What it does badly (or dangerously)
- Exact facts: figures, dates, quotes, addresses, laws, medication doses.
- Very recent events, because it learned up to a certain date.
- Long calculations done “in its head”; a calculator is better for those.
- Admitting on its own when it does not know.
An example prompt, good for your first test: “Explain how a chatbot works to an eighth-grade student, in at most 120 words, with an everyday analogy. At the end, tell me one thing a chatbot cannot do well.”
In the European Union, the official page about the AI Act says that people should be made aware when they interact with an AI system such as a chatbot. If a service answers you automatically, you have the right to know it.
Key takeaways
- Generative AI creates new content based on patterns; it does not read answers from a book of facts.
- It can write falsehoods with great confidence, so a sure tone does not mean a correct answer.
- Use it for drafts, explanations, summaries and ideas; check the important facts yourself.
- Without displayed sources, treat the answer as a well-written guess.
Try it now
Replace what is in braces with your own details, then copy.
Check what you learned
Three questions. They count for nothing but you.