Chapter 1 — Reps

Ten reps to turn Week 1 from something you read into something you can do. They move from noticing the AI already in your life, to naming what each one is doing, to catching your first hallucination and finding where the judgment lives in your own work. No coding. No accounts required for most of them. The keyboard is the gym — do not just read these, run them.

Ground rules

• Type it yourself. Fill the worksheets and write the reflections in your own words. Do not have an AI answer these for you — the point is to notice your work and your judgment, and you cannot outsource that.
• Try everything. Where a rep says open a tool, click a thing, or test an answer, actually do it. A rep you only imagined is a rep you did not do.
• Predict before you measure. Before you look up an answer, sort a product, or run a task, write down what you expect. Then compare. The gap between your prediction and the result is the learning.
• AI usage note (the honesty habit): Any rep where you do use an AI tool ends with one honest line — what you asked, what it got wrong, and what you verified yourself. We start this habit in Week 1 and keep it for sixteen weeks. The human owns the verdict.
• Keep a reps.txt. One short written reflection per rep, in a single document (Google Doc, Word file, or a plain notes file). This is graded thinking, not busywork — and it becomes the raw material for your project.

Reps 1–3: See the AI you already live with

Rep 1 — Inventory your own workday

Open code/ai-in-my-work.txt and fill in Part A for real. List up to eight tools or features you touch in a normal week that use AI — spam filter, autocorrect, maps ETA, photo search, fraud alerts, recommendations, meeting captions, a website chatbot. Do it from memory first; only then glance at your phone to catch the ones you missed.

Reflection: How many did you count before checking your phone, versus after? What does that gap tell you about how invisible this technology has already become in your life?

Rep 2 — Name the type of each one

Go back to your Part A list and label each entry with one of the six types (predictive / generative / conversational / multimodal / agentic / physical), using code/ai-types-cheatsheet.txt. For any that honestly fit two buckets, write both and star the one you think is primary.

Reflection: Which single type shows up most in your week? Most people find that the AI they rely on is mostly predictive and quietly invisible, while the AI they talk about is generative. Is that true for you?

Rep 3 — Where does the judgment live?

Fill in Part B of code/ai-in-my-work.txt: the one task in your week where getting it wrong would cost the most, the worst outcome if an AI drafted it unchecked, and the name of the person who would have to answer for that outcome.

Reflection: Is the highest-stakes task also the one you would most want to hand to AI? Explain the tension. This is the spine rule — you own the verdict — landing on your actual job.

Reps 4–6: Learn to see the machine underneath

Rep 4 — Draw the nesting dolls

Without looking back at §1.2, draw the three nested boxes from memory — Artificial Intelligence containing Machine Learning containing Deep Learning — and write one plain sentence defining each. Then check yourself against the chapter and fix what you got wrong.

Reflection: Which of the three did you define least confidently? Rewrite that one definition so a smart twelve-year-old would understand it.

Rep 5 — Explain the Transformer to a skeptic

In no more than five sentences, explain to an imagined skeptical coworker why AI seemed to explode after 2022 — using the words Transformer, predict the next word, and ChatGPT (2022). No jargon they wouldn’t understand.

Reflection: Where did you feel the urge to hand-wave? That spot is the edge of your real understanding. Note it — we open that hood in Chapter 2.

Rep 6 — “AI is the new electricity” — poke your own analogy

Write the five-sentence version of why AI is a general-purpose technology like electricity or the Internet. Then, in two more sentences, argue the other side: name one important way AI is not like electricity.

Reflection: A good analogy has a seam. What is the seam in “AI is the new electricity”? (Hint: electricity never confidently made up a fact.)

Reps 7–9: The first taste of judgment

Rep 7 — Catch a confident wrong answer

You do not need an account for this. Pick a question you already know the true answer to cold — a fact about your town, your field, a hobby, your family history. Ask it of any free AI you can reach (a chatbot, a phone assistant, a search “AI overview”). Read the answer with a cold eye and mark anything even slightly off.

Reflection: Was the answer right, subtly wrong, or confidently wrong? Describe the tone — did the wrong parts sound any less certain than the right parts? End with your AI usage note.

Rep 8 — Sort ten products into the six types

Make a two-column list. In the left column, write ten AI products or features (mix easy and hard — spam filter, image generator, robot vacuum, a research agent, a photo-to-recipe app, a self-driving feature, a translation app, a fraud alert, a chatbot, a music generator). In the right, assign each a type. Then use the Type-of-AI Sorter widget below Chapter 1 to check yourself.

Reflection: Which product surprised you by belonging to a different type than you first guessed, or to more than one? What would you have to verify differently for each type?

Rep 9 — Match the tool to the task (a preview)

For each of these three tasks, write which type of AI you’d reach for and one sentence why: (a) “spot which invoices look fraudulent,” (b) “draft a warm thank-you note to a volunteer,” (c) “book the three cheapest flights and put them in a table.” Don’t overthink the tool name — get the type right.

Reflection: Which of the three would you be most nervous letting AI do without checking, and why? Tie your answer to the spine rule.

Rep 10: Run the lab

Rep 10 — Score yourself on the Type-of-AI Sorter

Use the Type-of-AI Sorter widget embedded below the chapter. Sort every product it offers, commit, and record your score. Then read every explanation — especially for the products that belong in two or three buckets at once.

Reflection: What was your score, and which products did you miss? Were your misses on the definition (you didn’t know the type) or on the overlap (it was two types and you picked the wrong primary)? What does that tell you about where to sharpen before Project 1?

Done? One Last Thing.

A miniature of Project 1, done in fifteen minutes with no accounts required. Pick one small, real, low-stakes writing task from your actual week — a two-line thank-you note, a short reminder to a coworker, a caption. Then:

1. Write it yourself first, by hand, and save it.
2. Ask any free AI you can reach to write the same thing from a one-line request. Save that too, unedited.
3. In three sentences, compare them: what did the AI do better, what did it get wrong or make generic, and what would you have to change before you’d actually send its version?

If you can do this honestly for one small task tonight, you can do Project 1. That is the whole habit in miniature: draft, compare, verify, decide — and keep your name on what goes out. End with your AI usage note.

Up next: Project 1 — Project P1: First Contact.