1. Foundations6 lessons
  2. Inside the Model5 lessons
  3. How Learning Works5 lessons

16 lessons · 3 topics · each one builds on the last

KIeinfach verstehen

KI einfach verstehen shows you how artificial intelligence actually works — clearly explained, never just skimming the surface. Whether neural networks are completely new to you or you already bring some technical background: the lessons meet you exactly where you are.

Read the first lessonBrowse all topics

New here? How this site worksWhy it exists

The path through all 16 lessons

Three topics, one continuous path

Every lesson is a complete, self-contained chapter — not a quick post. The topics build on each other, and the map always shows you where on the path you currently stand.

  • readable now
  • in preparation
  • read by you
Topic 16 lessons

Foundations

The concepts everything else builds on — you'll sort out what separates a program from a model and what's behind probabilities, vectors, and hardware.

  1. Program, Algorithm, Model Compared
  2. Input and Output: How a Function "Thinks"
  3. How Language Becomes Numbers: Tokenizer, IDs, Vocabulary
  4. Scalar, Vector, Matrix, Tensor: The Building Blocks of the Numbers
  5. Probability and Softmax: How a Model Decides
  6. Parameters, Training vs. Inference, Hardware: How a Model Runs
Explore this topic
Topic 25 lessons

Inside the Model

A prompt goes in, a token comes out — you follow everything that happens in between, step by step.

  1. What an AI Model Actually Is
  2. Tokenization Inside the Model: Sequences, Special Tokens, Context Window
  3. Embeddings: How a Number Becomes a Meaningful Vector
  4. Transformer Blocks and Attention: How Context Gets Mixed In
  5. Output Head: From the Last State to a Prediction
Explore this topic
Topic 35 lessons

How Learning Works

Why a model gets better: you'll see how it measures errors, traces improvements backward through its calculation steps, and adjusts its weights step by step.

  1. Text Becomes Many Practice Problems
  2. Forward Pass and Loss: How the Model Measures Its Own Error
  3. Backpropagation and Gradients: How the Model Knows What to Change
  4. Optimization: What Actually Changes the Weights
  5. Batch, Epoch, Step, Token Budget: How to Measure Training Progress
Explore this topic

Start here

Readable nowLesson 1 · Foundations

Program, Algorithm, Model Compared

Is a language model just a very complicated program? The first lesson clears up exactly that misconception — and hands you the three terms every later lesson builds on.

  • ≈ 13 minutes of reading
  • With links into the glossary
  • Also available in German
Read lesson 1