A prompt is not a question. It is a task, instructions, and context. The machine is still predicting, and your words are what it predicts from.
If you remember one thing: context is the part almost everyone skips and the part that helps most. Starve the machine and it predicts the most average document on the internet; feed it and it predicts something shaped like you.
| Part | It covers | Cover letter example |
|---|
| Task | ”what you want the AI to do” | Draft a cover letter for the job posting below |
| Instructions | ”how you want the AI to do it” | Under three hundred words, warm but direct, no stock phrases, two versions |
| Context | ”what you want the AI to know” | Who you are, your proudest wins, the posting pasted between triple quote marks |
| Habit | Why |
|---|
| Start simple, then build through conversation | You refine faster than you can pre-plan |
| Be specific | Vague guidance buys the most average answer |
| Say what to do, not what to avoid | Naming the forbidden thing puts those words into play |
| Ask for many options, not one | Variability is fuel; pick from a menu |
| Fence off pasted material | Triple quote marks keep the posting separate from the request |
| Start fresh when a thread turns to sludge | The whole conversation is input; ask it to summarize the useful parts as a single prompt and carry that into a clean thread |
| Ask the machine to draft your prompt | When writing the prompt feels hard, describe the goal instead |
| Line | Meaning |
|---|
| One answer is a sample, not a verdict | In most chat tools the machine picks among plausible next words, like weighted dice; ask again, rephrase, request a second version |
| Variability is fuel | Twenty ideas cost nothing and hand you a menu; generate plenty and pick the best |
| Your prompt cannot switch the dice off. It decides which game the dice are playing. | Tight instructions and concrete examples narrow the menu when you need consistency |
| Upgrade | What you do | When it earns its keep |
|---|
| Ask for steps (chain-of-thought) | Ask the machine to walk through its steps, with a reason for every judgment | Analytical tasks; the visible reasons expose a hidden yardstick, like extreme-language scoring in the Boston Globe comment example |
| Show examples (few-shot) | Give two or three labeled samples of what you want | When describing what you want is hard but showing it is easy |
Aging note: when the course was filmed in 2024, think step by step was a measurable trick. Many newer systems, often described as reasoning models, now do much of that on their own. Visible steps still let you check the work, but the written explanation is itself output to check.
Roles worth borrowing: mentor, critic, tutor, student, interviewer, simulator.
| Setup tip | Why |
|---|
| Open with the words “you are” | State an identity rather than asking it to pretend |
| Say you will speak turn by turn | Otherwise it may write the entire two-sided conversation itself |
| Give the character a spine | Practice negotiators cave instantly unless told to be tough and hold certain lines |
| End with an out-of-role critique | Ask it to step out of the role and critique your performance; that turns a simulation into coaching |
| Pitfall | Correction |
|---|
| Judging the machine on a one-line prompt | Give a real task the three-part treatment before deciding what the tool can do |
| Telling it what not to do | Every forbidden word you name is a word you placed on the table |
| Living in one endless thread | Old dead ends leak into new replies; summarize and start clean |
| Believing the character | A persona changes style and stance, not knowledge; a practice negotiator does not know any company’s internal salary bands. Rehearse with the character, verify with sources |
| Line | Meaning |
|---|
| ”having the first result of your first Google search color what you think of the internet” | One of the course’s teachers on judging the technology by a single weak reply |
| ”giving an AI personality” | The course’s name for the persona move |
| Ask well, and good answers stop being luck. | The lesson’s closing line: prompting is clear thinking made visible |
