AI Study Protocol
Principle β Prepare with AI. Prove it in person, without devices.
Using AI as a tutor is the assignment. The goal is the ability to learn on your own with AI.1. Allowed β preparation (use AI as a tutor, actively)
β’
Ask for re-explanations, alternative proofs, counterexamples
β’
Generate practice problems β solve and check them
β’
Verify claims numerically with NumPy
β’
Polish and rehearse your English presentation
β’
Any product, personal account, free tier is fine (Claude / ChatGPT / Gemini / NotebookLM β¦). No institutional license.
2. Not allowed β live
β’
During your presentation, no AI; Q&A answered live with no AI
β’
Midterm & final exams (10/21 Β· 12/16): writing instruments only
β’
The weekly pset scan must be your own handwriting β no typed or AI-generated solutions
3. Verification ladder β at least one per AI-derived claim
Step | Method | Tag |
1 | Textbook page | [LALFD p.71, I.9] |
2 | Lecture timestamp | [Lec 7, 14:20] |
3 | NumPy check | [numpy: 3 trials, atol=1e-8] |
4 | Counterexample | [counterex: A rank-deficient] |
4. AI-use written interview in exams (announced in advance)
Both exams include a written interview: how did you use AI as a tutor to understand this material β for understanding, not homework or problem-solving.β’
Written in the exam room, no devices, ~200β300 words from your own learning.
β’
Graded on sincerity, specificity, and a verifying attitude (not right/wrong).
β’
Announced from day one β so keep notes on how you use AI while you study.
5. Per-session submission (participation) β due D-1 24:00
Before every session, submit these three items on the LMS by the midnight before class:
1. Muddiest Point β the one thing still unclear after studying.
2. Hand-solved pset scan β the assigned pset problems worked out by hand (photo/scan, PDF). Handwriting only.
3. AI-use note (2β4 sentences) β whether AI tools helped your study, and how AI helped you overcome what you could not understand on your own.
Keeping an honest record of how AI helped you (not whether it gave you the answer) is exactly the habit this course grades.AI νμ΅ νλ‘ν μ½
μμΉ β μ€λΉλ AIμ, μ¦λͺ
μ μ¬λ μμμ κΈ°κΈ° μμ΄.
AIλ₯Ό νν°λ‘ μ°λ κ²μ μ΄ κ³Όλͺ©μ κ³Όμ λ€. λͺ©νλ AIλ‘ μ€μ€λ‘ νμ΅νλ λ₯λ ₯μ΄λ€.1. νμ© β μ€λΉΒ·νμ΅ λ¨κ³ (AI νν°λ‘ μ κ·Ή νμ©)
β’
κ°λ
μ λ€λ₯Έ λ°©μμΌλ‘ μ¬μ€λͺ
λ°κΈ°
β’
λ€λ₯Έ μ¦λͺ
κ²½λ‘Β·λ°λ‘ μμ²
β’
μ°μ΅λ¬Έμ μμ± β νκ³ κ²μ°
β’
NumPyλ‘ λͺ
μ μμΉ κ²μ¦
β’
μμ΄ λ°ν μ€ν¬λ¦½νΈ λ€λ¬κΈ°Β·λ¦¬νμ€
β’
μ΄λ€ μ νμ΄λ κ°μΈ κ³μ Β·λ¬΄λ£ ν°μ΄λ‘ 무방 (Claude/ChatGPT/Gemini/NotebookLM λ±). κΈ°κ΄ λΌμ΄μ μ€ μμ.
2. κΈμ§ β λΌμ΄λΈ
β’
λ°ν μ€μλ AI μμ΄; Q&A μ¦λ΅λ AI μμ΄
β’
μ€κ°Β·κΈ°λ§ μν(10/21Β·12/16): ν기ꡬλ§
β’
λ§€μ£Ό pset μ€μΊμ λ³ΈμΈ μνλ§ β νμ΄νΒ·AI μμ± νμ΄ κΈμ§
3. κ²μ¦ μ¬λ€λ¦¬ β AI μ λ μ£Όμ₯μλ μ΅μ 1κ°
λ¨κ³ | λ°©λ² | νκΈ° μ |
1 | κ΅μ¬ μͺ½μ | [LALFD p.71, I.9] |
2 | κ°μ νμμ€ν¬ν | [Lec 7, 14:20] |
3 | NumPy μμΉ κ²μ¦ | [numpy: 3 trials, atol=1e-8] |
4 | λ°λ‘ νμ | [counterex: A rank-deficient] |
4. μνμ AI νμ© μλ©΄ μΈν°λ·° (미리 μλ΄)
μ€κ°Β·κΈ°λ§ μνμλ βμ΄ λ²μλ₯Ό νμ΅νλ©° AIλ₯Ό νν°λ‘ μ΄λ»κ² μΌλκ°β λ₯Ό μμ νλ μΈν°λ·°κ° ν¬ν¨λλ€. μμ λνΒ·λ¬Έμ νμ΄κ° μλλΌ μ΄ν΄λ₯Ό λλ μ©λλ₯Ό λλμ보λ κ²μ΄λ€.β’
μνμ₯μμ κΈ°κΈ° μμ΄, λ³ΈμΈ νμ΅ κ²½νμ 200~300λ¨μ΄λ‘.
β’
μ±μ : μ±μ€μ±Β·κ΅¬μ²΄μ±Β·κ²μ¦ νλ (μ μ€ μλ).
β’
νκΈ° μ΄λΆν° μλ΄ β κ·Έλ¬λ νμ AIλ₯Ό μ΄λ»κ² μ°λμ§ κΈ°λ‘ν΄ λλ©΄ μ 리νλ€.
5. λ§€ μΈμ μ μΆλ¬Ό (μ°Έμ¬) β D-1 24:00κΉμ§
λ§€ μΈμ
μ λ μμ κΉμ§ μλ 3μ’
μ LMSμ μ μΆ:
1. Muddiest Point β μ¬μ νμ΅ ν κ°μ₯ λ§ν μ 1κ°.
2. pset μνμ΄ μ€μΊ β μ§μ pset λ¬Ένμ μμΌλ‘ μ§μ νΌ μ¬μ§/μ€μΊ(PDF). μνλ§.
3. AI νμ© μμ (2~4λ¬Έμ₯) β AI λꡬμ λμμ λ°μλμ§, κ·Έλ¦¬κ³ μ΄ν΄νμ§ λͺ»ν κ²μ AI λμμΌλ‘ μ΄λ»κ² 극볡νλμ§.
βμ λ΅μ λ°μλκ°βκ° μλλΌ βAIκ° μ΄ν΄λ₯Ό μ΄λ»κ² λμλκ°βλ₯Ό μ μ§νκ² κΈ°λ‘νλ μ΅κ΄ β μ΄ κ³Όλͺ©μ΄ νκ°νλ κ²μ΄ λ°λ‘ κ·Έκ²μ΄λ€.

