Skip to content

MISC AI Learning - Template

AI-Era Universal Learning Template v1

Overview (the entire flow at a glance)

Step Name What You Do Time
0 Classify What type is this task? (Production / Understanding / Judgment / Integration) 0.5
1 Define Ask the right 3 questions for your type 0.5
2 Draft with AI Get different materials from AI based on your type 2
3 Verify Confirm AI's output is correct / confirm you actually understand 3
4 Vary Change one condition to prove you're not just memorizing 3
5 Integrate Place it in a larger system, make it reusable (optional) 0
6 Compress Write a card, save to learning-log.txt 1

Each step adapts based on your Step 0 classification.


Comparison: Traditional Homework Flow vs This Template

Traditional Mapped Step AI's Role Your Focus
Get assignment (input)
Understand problem Step 0 Classify Can ask AI to clarify Decide what type
Step 1 Define Ask the right questions
Do the work Step 2 Draft ⚡ AI produces first draft You set direction, AI executes
Check answers Step 3 Verify ⭐ AI helps test & quiz you You think deeply
Review mistakes Step 4 Vary ⭐ AI can help analyze You modify & reflect
Step 6 Compress You write the card

Key Insight

⚡ = Where AI saves you the most time (reinvest that time into ⭐)

⭐ = Where real learning happens (deep thinking + AI-assisted dialogue)

Traditional: Most time spent on "doing the work"

AI-era: Doing the work is faster → shift time to "verify + vary"


Core Principle

Before AI, learning forced "output" and "ability" to be tied together. After AI, output is cheap → learning must shift to defining problems, verifying, varying, integrating.


Step 0 | Classify + Prioritize

Ask yourself: What's my gut reaction?

Gut Reaction → Type
"Just get it done" 🔵 Production
"I don't understand this" 🟢 Understanding
"I'm not sure which to choose" 🟠 Judgment
"I know the parts but can't connect them" 🟣 Integration

Most tasks are 1 primary + 1 secondary. Just identify the primary.

Time allocation (AI : You):

Type AI : You Rationale
🔵 Production 7 : 3 AI is strongest here; you just confirm and fix
🟢 Understanding 4 : 6 AI explains; you internalize
🟠 Judgment 3 : 7 AI lists options; you decide
🟣 Integration 5 : 5 AI writes parts; you connect them

Context reference:

Scene Most Common You Should Train
Student work Production + Understanding Understanding + Judgment
Work tasks Production + Integration Judgment + Integration
Side projects Integration Integration + Judgment

Step 1 | Define

Based on your classification, ask the corresponding 3 questions:

Define the delivery boundary

  1. What to deliver? (format, audience)
  2. What NOT to do? (scope)
  3. Minimum acceptable standard? (good enough to submit)

Define the confusion boundary

  1. Where am I stuck? (specific breakpoint)
  2. What do I already understand? (starting point)
  3. How do I know I understand? → Default to L3 (see below)

Define the decision framework

  1. What are my options? (list 2–3)
  2. What criteria? (max 3, ranked by priority)
  3. When must I decide?

Define interfaces and order

  1. How many parts? (one sentence each)
  2. How do they connect? (who feeds whom)
  3. Which to connect first? (minimum viable path)

5 Levels of Understanding

For the Understanding type's "How do I know I understand?"

Level Self-Test Pass Criterion
L1 Recognize See it, know what it is Can name it
L2 Explain Explain without looking 30 seconds, no stumbling
L3 Predict "If ___ happens, then ___ would..." Can answer ← use this by default
L4 Modify Change one condition Know how the result changes
L5 Rebuild Without looking Can reconstruct the core structure

Context:

Scene Target Level
Exam concepts L2–L3
Homework L3–L4
Using others' code L3
Designing yourself L4–L5
Job interviews L2–L3

Step 2 | Draft with AI

Based on your classification, ask AI for different things:

  • Minimum viable version, just make it work
  • Give it to me in modules
  • Tell me what you assumed
  • I understand ___, but I'm stuck on ___
  • Walk me through a concrete example (input → process → output)
  • Don't give me definitions, give me the process
  • I'm choosing between A / B / C
  • Compare using my criteria ①②③
  • Give me a table
  • I have these parts: ___
  • Define each part's input/output
  • Suggest assembly order (which connection to verify first)

Checkpoint

After getting AI's output, do you know what to do next?

Yes → Continue to Step 3

No → Your Step 1 definition wasn't specific enough. Go back and rewrite it.


Step 3 | Verify

  1. Does it run?
  2. Does it break at boundaries? (different input / edge cases)
  3. Are AI's assumptions valid in my context?

Leave behind: test cases you ran

  1. L3 self-test: "If ___ happens, then what?"
  2. Ask AI to quiz you with 2–3 variation questions
  3. Explain it in your own words for 30 seconds

Leave behind: your own 3–5 line explanation (not copied from AI)

  1. Under what circumstances would I change my decision?
  2. Can I explain my choice in one sentence?
  3. Worst case scenario — can I live with it?

Leave behind: one-sentence decision + one-sentence rationale

  1. Is the data format consistent at every connection point?
  2. If I remove one part, can the error be detected?
  3. Does the minimum viable path work end-to-end?

Leave behind: status of each connection point (pass / fail / issue)


Step 4 | Vary

Change format / change scale / add a small feature

→ If it still works after the change = you truly understand the code

Same concept, different context / same algorithm, different input

→ Complete it without looking at the answer = you understand the principle, not just the example

Time changes / requirements change / resources change

→ Can articulate whether your decision changes and why = real judgment

Swap implementation / swap protocol / replace with a stub

→ Know which interfaces to update and it still works = you understand every connection

Short on time?

At minimum, spend 30 seconds thinking: "If I changed ___, what would happen?"


Step 5 | Integrate

Core question: Where does this thing live after it's done?

  1. How does this relate to other assignments in this course?
  2. What can I reuse next time I face a similar problem?
  1. How will others use what I built?
  2. How do we debug it when it breaks?
  3. Will I understand this in 3 months?
  1. Can I explain this project in 10 minutes?
  2. Can I make changes in 30 minutes if I come back in 6 months?

Skip rule

If you won't revisit this in 3 months → skip.


Step 6 | Compress

Write one card, no more than 6 lines:

Date:
Type: (Production / Understanding / Judgment / Integration)
What I did: (1 sentence)
Key principles: (3-5 terms)
Variation / learned: (1 sentence)
Reusable next time: (template / checklist / module / decision framework)

Store all cards in a single file (learning-log.txt). Keep it searchable.