Start here: LLM literacy without code

This route is for managers, analysts, writers, teachers, students and policy and legal people who need to know what AI systems can and cannot do. You need curiosity and about six hours a week for nine weeks, and no maths beyond percentages.

In Daveth Pro (AI Engineering)

What this page answers

Can I learn AI without coding?

Yes. This route reads the intuition sections of each chapter, tries every interactive demo and answers the prediction questions, with no maths beyond percentages. Read it in “The route”

Who is AI for non-programmers for?

Managers deciding what to automate, analysts and writers using AI every day, teachers and students, policy and legal people, and anyone who would like to stop nodding along. Read it in “Who this is for”

Who this is for

Managers deciding what to automate, analysts and writers using AI every day, teachers and students, policy and legal people who need to know what these systems can and can't do, and anyone who'd like to stop nodding along. You need curiosity and about six hours a week for nine weeks. You don't need maths beyond percentages; where a formula appears in a lesson, the words around it say what it means.

The route

Each chapter page has three depth levels, marked by the small bars next to every section heading: one bar is Intuition, two is How it runs, three is Under the hood. On this route you read the one-bar sections of each chapter, try every interactive demo, and answer the “Predict first” questions, which need no code. Skip anything with two or three bars unless you're enjoying it. How to read the list: weeks are a suggested pace at about six hours a week, and the Track E lessons slot in where they make most sense.

Week 1

01 Predicting the next word · the one idea behind every chatbot Chapter

E1 How a chatbot actually works · Track E Lesson

Week 2

02 Meaning as numbers · how words become points in space Chapter

03 Uncertainty and change · probability and learning, without the algebra Chapter

Week 3

04 Teaching models from data · what “training” means, and how to judge a model Chapter

E2 Why models make things up · Track E Lesson

Week 4

05 Backprop from scratch · Level 1 only: how a network improves itself Chapter

06 Text to tokens to vectors · why models count tokens, not words Chapter

Week 5

07 Building mini_lm v1 · watch a small language model being built and trained Chapter

E3 Prompting from the mechanism · Track E Lesson

Week 6

08 The right assumptions · why some designs learn faster Chapter

8.4 Self-attention · the idea that made modern AI Chapter

09 Mini-GPT, block by block · a small GPT, end to end Chapter

Week 7

10 Scaling laws, measured · why companies build ever-bigger models Chapter

11 Base model to assistant · how a text predictor becomes a helpful assistant Chapter

E4 Using AI at work safely · Track E Lesson

Week 8