Your Study Routine

One routine, not five techniques

Five lessons of individual tactics are worth very little if you have to decide which one to use every time you sit down. What actually changes how you study is a default: something you do without thinking, in the same order, whatever the subject.

So here is the routine the whole course has been building towards. It works for a module you are behind on, a paper you have to present on Thursday, a topic you need for a job, or anything else you have to genuinely learn rather than skim. It runs in five steps and it maps exactly onto the four jobs from Lesson 1.

Step 1: Map the territory before you read anything

Fifteen minutes, before you open a single source. You are working out what this topic is made of, what the technical vocabulary is, and what the arguments are, so that the reading makes sense when you get to it.

This is where AI earns its place most obviously. Ask what specialists call the thing you are describing in everyday words, what the main positions are, and what the beginner mistakes look like. Do not ask for reading. You are collecting search terms, not sources.

I need to learn about [topic] for [essay / exam / a meeting / a report].
My level: [where you are]. Time I have: [be honest].
Give me:

The technical terms specialists use for this
The main positions or schools of thought, and who disagrees with whom
The five things I must understand before the rest makes sense
The mistakes beginners typically make on this topic
Eight search queries to run in a library database or search engine
No references, no reading lists. I will find the sources myself.

Step 2: Find real sources yourself

Take the search terms into your library, your reading list, or an academic search engine. Get real documents. Open every one you intend to cite.

The rule from Lesson 2 does not bend here, and it is the one that ends academic careers when it does: never cite anything you have not opened.

Step 3: Understand it, one sticking point at a time

Read. When you get stuck, do not reread the paragraph a fourth time. Paste it, say what level you are at and what specifically is not landing, and ask for the same idea three ways: plainly, as an analogy, and as a worked example. Then ask where the analogy breaks.

Check the explanation against the source before you accept it. Fluent is not correct.

Step 4: Turn it into questions, not notes

This is the step people skip, and it is the one that decides whether any of the last three hours survives the week.

Write four sentences from memory, without looking. Have them corrected against the source. Then generate ten practice questions with no answers, attempt them badly, and have your attempt marked.

Stage 1. Here is the material and here is my own summary, written
from memory without looking. Tell me what is factually wrong, what
important thing I left out, and where I have overstated the claim.
Stage 2. Then write me 10 practice questions on this material.
Questions only, no answers. Mixed difficulty, easiest first.
At least three should ask why rather than what, and at least two
should apply the idea to a case not in the material.
I will send my answers back for marking. Be blunt throughout.
Material:
[paste it]
My summary from memory:
[paste it]

Step 5: Produce the work yourself, with feedback

Write it. Your words, your argument, your structure.

Then use AI the way a good supervisor would be used: ask what you have not supported, where your reasoning jumps, what the strongest objection is, and where you are describing rather than arguing. Do not ask it to rewrite anything, and do not let it hand you text you could paste in.

And check your institution's rules before you do any of this, because they vary and yours are the only ones that count.

The checklist

Print it, copy it into your notes app, or write it on the inside cover of your notebook. It is short on purpose.

If any row is blank, that is the next hour of work, and it is a more valuable hour than another read-through.

Where to go next

Three courses on this site build directly on what you have just done.

Prompt Engineering goes deeper on getting better answers out of any tool. Most of the prompts in this course follow patterns from it: state your level, state what you want, state what you do not want.

AI Safety, Privacy and Verification covers checking claims properly, working out what is safe to paste into a tool, and spotting fakes. Everything in Lesson 2 about verifying sources has a fuller treatment there.

AI Productivity at Work is the version of this routine for people whose deadlines come from a manager rather than a marking scheme.