Artificial intelligence
🧠 How AI Works: The Kinds of AI and What They Are For
Learn what the word AI covers, how machines learn from data, how chatbots and image generators actually produce their output, and why they get things wrong. Finish able to tell the kinds of AI apart and read a model announcement without getting fooled.
- 1What "AI" Means, and How We Got HereThe word AI covers seventy years of very different machines. Learn the main eras, from hand-written rules to the 2017 transformer and the 2022 ChatGPT launch, and why each one mattered.14 min
- 2The Kinds of AI and What Each Is ForA field guide to ten kinds of AI, from rule-based systems to agents that use tools, with what each one does well, how it fails, and where you already meet it.16 min
- 3How Machine Learning LearnsData, features, training, loss, overfitting and test sets, worked through a tiny example you can do with a pencil.17 min
- 4Neural Networks and Deep Learning in Plain WordsWhat an artificial neuron is, how weights and layers stack up, and how gradient descent nudges billions of numbers toward better answers.16 min
- 5How Large Language Models WorkTokens, next-token prediction, attention, pretraining, fine-tuning and RLHF, the context window and temperature, each with a worked example.19 min
- 6Why Chatbots Make Things Up, and What Fixes HelpWhere hallucinations come from, how retrieval and search grounding cut them down, and what reasoning models and tool use add, with the limits of each.18 min
- 7How Image, Audio and Video Generators WorkDiffusion from noise, text encoders that connect words to pictures, latent space, video as moving diffusion, and how text-to-speech and voice cloning work.17 min
- 8Narrow vs General, Open vs Closed, Local vs Cloud, and AgentsFour distinctions that come up in nearly every AI announcement and debate: narrow versus general AI, open versus closed weights, running a model on your own machine versus in the cloud, and what makes something an agent.17 min
- 9Who Makes What: Model Families as of October 2026, and a Working GlossaryA map of the major AI companies and their model families as of October 2026, how to read a model name and a launch announcement, and a glossary of the terms from this module.18 min