AI basics, a course in five parts

5 articles, read in order

• September 28, 2026 · 8 min read

  Part 1: What Is AI? Machine Learning, Deep Learning and LLMs Explained Simply

  The four words everyone mixes up, sorted out in plain language. What artificial intelligence, machine learning, deep learning and large language models actually mean, how they nest inside each other, and where ChatGPT fits.

  machine learning deep learning llm

• September 28, 2026 · 7 min read

  Part 2: How a Neural Network Learns: Weights, Loss and Gradient Descent by Hand

  Part two of our AI basics series. We build a neural network with one weight, train it by hand on three examples, and meet every idea that scales up to GPT: weights, loss functions, gradients, learning rate, epochs and overfitting.

  neural networks gradient descent training

• September 28, 2026 · 7 min read

  Part 3: How Language Models Read Text: Tokens and Embeddings Explained

  Part three of our AI basics series. Neural networks only multiply numbers, so how does one read a sentence? Tokens, tokenizers, why models are bad at counting letters, embeddings, and why 'king minus man plus woman' lands near 'queen'.

  tokens embeddings tokenizer

• September 28, 2026 · 7 min read

  Part 4: Transformers and Attention Explained Without the Maths

  Part four of our AI basics series. What the T in GPT stands for, what attention actually does, why it replaced older designs, why models have a context window, and why long prompts cost more than you expect.

  transformers attention context window

• September 28, 2026 · 8 min read

  Part 5: From Next Word Predictor to Chatbot: Pretraining, Fine Tuning and RLHF

  Part five of our AI basics series. How a raw text predictor becomes ChatGPT: pretraining, instruction tuning, reinforcement learning from human feedback, plus what temperature does, why models hallucinate, and what a system prompt is.

  chatgpt rlhf fine tuning training