1How Computers Learn by TryingThis guide assumes nothing. It starts with something everyone already understands — learning by trying — and builds up to explain one of the most important ideas in modern AI. No maths required. Every term is explained when it first appears.The basics 2How AI Learns to Be HelpfulThis guide assumes you have read the first guide in this series (or already know the basics of reinforcement learning). It explains the technology that transformed AI from “impressively fluent but often useless” to “genuinely helpful” — without using any mathematics.RLHF, explained simply 3When AI Finds LoopholesThis guide explains one of the most important ways AI systems can go wrong — not through malice, not through a coding error, but through perfectly successful optimisation of the wrong goal. Every term is explained when it appears. No prior knowledge of AI is assumed.How AI goes wrong 4How AI Agents Make DecisionsAI agents decide by turning goals, context, predictions, rules, and feedback into actions under uncertainty.Agents and decisions 5AI That Never Stops LearningMost software does the same thing today that it did last week. Some AI systems are different: they keep learning from real-world experience long after they are deployed. This guide explains what that means, why it matters, and what everyone — not just engineers — deserves to understand about it.Ongoing learning