Learning path

AI literacy basics

From the technical foundations to the chatbot in everyday use: what an AI model is made of, how language becomes numbers, how a model becomes an assistant, why it can be convincingly wrong, and how you check its answers.

For anyone who uses AI at work or in everyday life and wants to understand what happens technically. No prior knowledge needed.

  • 17 lessons
  • ≈ 229 min reading time (about 4 h)
  • 93 questions in recall moments

Reading time is estimated from the text length (200 words per minute). Recall moments and interactive demos come on top.

0 of 17 lessons completedStart the learning path

Your progress is stored only in your browser, on this device. There is no account. Transfer it to another device

Learning goals

After this learning path you can …

  • Tell program, algorithm and AI model apart, and explain why a model is not made of written-down rules but of numbers that are set during training.
  • Describe what an AI model receives as input, why its output is usually a list of ratings, and how an answer is built from it piece by piece.
  • Explain how a tokenizer splits text into pieces from a fixed list, how it learns these pieces by counting, and why the number of tokens is not the number of words.
  • Explain how tokens become numbers and back, why a number means nothing by itself, why tokenizer and model belong together, and how many tokens a model processes at once.
  • Place scalar, vector, matrix and tensor, and explain why a chat message or a photo is a block of numbers to a model.
  • Explain how a language model turns scores into probabilities, why it does not always pick the obvious option, and what temperature changes about that.
  • Describe what is inside a finished model, what is fixed before training, how training gets the direction for each parameter from the error, and why training costs more than inference.
  • Estimate how much memory a model needs and explain why space decides first, what quantization saves, and why training needs more memory.
  • Explain why good prediction needs rules rather than copies, how a task that never existed before can still succeed, and where verbatim memorization and truly new tasks hit limits.
  • Explain why a base model continues questions instead of answering them, what post-training changes about that, and where a chatbot gets its date and its instructions from.
  • Explain why intermediate steps a model writes as text before answering make some tasks solvable, who decides whether and how long it thinks, and what a displayed chain of thought does not show.
  • Explain why a chatbot would rather guess than say “I don’t know”, why asking the same bot again is no check, and why an invented source looks real.
  • Name the four sources a chatbot draws on in a conversation (weights, context, tools, saved notes) and work out from them how reliable a statement is.
  • Explain how an agent uses tools in a loop although the model only writes text, who decides whether a call is run, and why the end of a run says nothing about the result.
  • Match claims from the debate on whether language models understand to the findings they rest on, and explain why a fluent answer speaks neither for nor against understanding.
  • Explain at which points a text is legally affected in training, which lever works for which text, and why an opt-out only affects future collection (dated, no legal advice).
  • Rate a chatbot answer by its risk and choose the check that fits the possible error, instead of asking the bot itself whether it is right.

Order

The lessons on this path

Every lesson also stands on its own. In this order, each builds on terms the previous one explained. A lesson counts as completed once you have read it to the end and answered every question of its recall moment once. Whether right or wrong does not matter for that.

  1. Program, Algorithm, Model Compared

    How do program, algorithm and model differ? Classic software follows rules a person writes down. An AI model consists of a fixed calculation and numbers set during training.

    CompletedRead≈ 16 minRecall moment: 0 of 6 questions
  2. Input and Output: What a Model Does with an Input

    What goes into an AI model, and what comes out? From spam filters to chatbots, the output is often a list of scores.

    CompletedRead≈ 14 minRecall moment: 0 of 6 questions
  3. Tokenizers: How Text Breaks into Tokens

    Shows how a tokenizer splits text into pieces that it learned earlier by counting, and why the number of tokens is not the number of words.

    CompletedRead≈ 16 minRecall moment: 0 of 8 questions
  4. Token IDs: What a Number Means

    Shows what a token ID is: a number that only has meaning within its tokenizer’s vocabulary. It also explains why each model must use the tokenizer it was trained with, and why the number itself reveals nothing about the piece.

    CompletedRead≈ 7 minRecall moment: 0 of 5 questions
  5. Scalar, Vector, Matrix, Tensor: the Building Blocks of Numbers

    From the weather app to the chatbot: how number, list, table and stack of tables relate, and why your chat is exactly such a block of numbers for a model.

    CompletedRead≈ 14 minRecall moment: 0 of 7 questions
  6. Probability and Softmax: How a Model Decides

    How a language model turns its scores into shares, why it sometimes picks the obvious choice and sometimes something else, and how temperature changes that.

    CompletedRead≈ 16 minRecall moment: 0 of 8 questions
  7. Parameters, Training and Use: How a Model Learns

    What is inside a model file, what people decide before training, how training uses the error to find each fader’s direction and why learning costs more than using.

    CompletedRead≈ 16 minRecall moment: 0 of 6 questions
  8. Model Size and Hardware: Where a Model Fits

    How much memory an AI model needs, why space, not speed, is the first limit, how quantization shrinks models and why training needs far more memory still.

    CompletedRead≈ 14 minRecall moment: 0 of 4 questions
  9. Why Prediction Is More Than Parroting

    How a model that only predicts the next token can write an email that never existed before: why good prediction requires rules rather than copies, where verbatim memorization still happens and where new tasks fail.

    CompletedRead≈ 16 minRecall moment: 0 of 6 questions
  10. From Text Predictor to Assistant: What Post-Training Makes of a Model

    Why a base model continues questions instead of answering them, what post-training changes, and where a chatbot gets its date and its instructions.

    CompletedRead≈ 13 minRecall moment: 0 of 4 questions
  11. How a Model "Thinks": Intermediate Steps It Writes Itself

    Why a model writes intermediate steps as text before its answer, why that makes some tasks solvable, and who decides whether and how long it thinks.

    CompletedRead≈ 16 minRecall moment: 0 of 6 questions
  12. Why AI Is Convincingly Wrong: When Guessing Pays Off and a Confident Tone Proves Nothing

    Why a chatbot would rather give a wrong answer than "I don't know", why follow-up questions often talk it out of a right answer, and why invented sources look real.

    CompletedRead≈ 11 minRecall moment: 0 of 5 questions
  13. What a Chatbot Knows in a Conversation: Four Sources, One Context Window

    Where an answer in a chat can come from: from the weights set by training, from the text in the context window, from tools and from saved memories, and what that means for how far you can rely on it.

    CompletedRead≈ 12 minRecall moment: 0 of 4 questions
  14. From Chatbot to Agent: How AI Works through a Task

    A model only writes text and runs nothing itself. How it becomes an agent that uses tools in a loop, who does what in that loop, and why the end of a run says nothing about the result.

    CompletedRead≈ 12 minRecall moment: 0 of 5 questions
  15. Does AI Understand What It Says?

    Experts disagree on whether language models understand. What the findings show, what they do not show, and why a fluent answer proves neither side.

    CompletedRead≈ 13 minRecall moment: 0 of 5 questions
  16. Training Data: Copyright and Privacy

    Which texts may be used for training, how a rights reservation works, why copying raises the first legal question, and what may remain in the model. As of 12 Oct 2026; not legal advice.

    CompletedRead≈ 13 minRecall moment: 0 of 4 questions
  17. Checking Answers: Using AI Safely in Everyday Life

    When a chatbot answer has to be checked, why asking the same bot again is not enough, and which checking step fits which kind of error.

    CompletedRead≈ 10 minRecall moment: 0 of 4 questions

To be honest

What this learning path does not cover

The path explains how AI models work technically. Using AI safely at work takes more, and this is not covered here:

  • obligations under the EU AI Act and legal advice; copyright and data protection appear only for training data, dated and without legal advice
  • risks and limits of specific AI systems in your own work context
  • how to operate particular tools or write prompts

At the end

Your learning record

Once you have read every lesson and done its recall moment, you can create a learning record: with your name, the date, the learning goals and the scope, to print or save as a PDF. It is a self-declaration, not verified, and deliberately carries no score. It can be part of your documentation of AI literacy measures (Art. 4 EU AI Act).

Go to the learning record

Want to use the learning path in teaching or training? Notes for educators & trainers