What Is a Large Language Model? (LLMs Explained From Zero)

Try it yourself — the full written explainer and an interactive next-word predictor are here: https://unrote.com/ai/what-is-an-llm/ A large language model sounds mysterious, but underneath it's one simple idea: a giant math function that, given some text, predicts the most likely next chunk of text — then feeds its own guess back in and does it again. That loop, repeated, is the whole engine. This is a build-from-zero explainer. No jargon assumed. We start with the autocomplete on your phone, watch a model generate a sentence one word at a time with real probabilities, and end up understanding why it can sound brilliant and still be confidently wrong. What we cover: - Why an LLM is really just autocomplete, scaled up enormously - Predict, append, repeat — how it writes one token at a time - Why it works in tokens (chunks), not whole words - Where the skill comes from: training on a huge pile of text - Why "Large" matters — billions of parameters - Why it doesn't actually "know" anything, and why that causes hallucinations - Why doing one simple thing well, at scale, ends up looking like intelligence Chapters: 0:00 What a large language model is 0:26 The whole thing in one sentence 0:44 Read the name backwards: L, L, M 1:03 It's autocomplete you already use 1:42 Watch it predict, word by word 2:36 It works in tokens, not words 2:59 Where the skill comes from: training 3:45 Why "Large" matters 4:10 It doesn't know — it predicts 4:33 So why does it feel intelligent? 4:58 Where this goes next 5:21 Recap This is the first video in an AI series on Unrote — modern AI explained from zero, one idea per page. Up next: What Are Tokens? Unrote. Understand it, don't memorize it.