Evidence-Based Daily Schedule for AI Learners: 3 Practical Examples and AI Prompt Collection

Overview

In the previous articles “Evidence-Based Effective Learning Methods” and “Effective Learning Methods Using AI Tools”, we explained methods to maximize learning efficiency based on 14 peer-reviewed research papers.

However, questions remain: “How should I specifically spend my day?” and “How should I instruct AI to create an optimal schedule?”

This article integrates research findings from the two previous articles to provide:

1. Prompt Collection for Having AI Create Your Schedule - Practical prompt examples you can use immediately
2. A Day in the Life of 3 Different Personas - Specific schedules for full-time engineers, students, and freelancers
3. AI Personal Assistant Settings - ChatGPT Custom Instructions and Claude Projects configuration examples
4. AI Utilization in Weekly Reviews - How to conduct reflection and improvement together with AI

Each schedule includes scientific evidence (references to research papers) explaining why each activity is performed at that timing.

Target Audience: IT engineers, students, and freelancers learning with AI tools Reading Time: 30 minutes Related Articles:

• Evidence-Based Effective Learning Methods
• Effective Learning Methods Using AI Tools

---

Setting Up AI Personal Learning Assistant

First, set up AI as your personal assistant for effective schedule creation and learning support.

ChatGPT Custom Instructions Example

Go to ChatGPT Settings → Custom Instructions and configure the following:

“What would you like ChatGPT to know about you to provide better responses?”

I'm an IT engineer (backend development) learning new technologies while using AI tools.

**Learning Style:**
- Value scientifically-based learning methods
- Practice AI Learning Pomodoro (15min learning + 5min break)
- Prefer interactive learning (Socratic dialogue)
- 70/30 Rule: 70% think for myself, 30% rely on AI

**Current Learning Goals:**
- Short-term (1-3 months): Master Rust basics
- Medium-term (3-6 months): Implement performance-critical features in Rust
- Long-term (6-12 months): Improve systems programming skills

**Learning Environment:**
- Remote work
- Learning time: Weekday mornings 7:00-8:30, evenings 19:00-20:30
- Weekends: Concentrated learning Saturday morning

**Research Findings I Value (from previous articles):**
1. AI Learning Pomodoro (breaks and memory consolidation)
2. Interactive learning (Socratic dialogue)
3. Leveraging curiosity and prior knowledge (think before asking AI)
4. Bug-driven learning (turn surprises into learning opportunities)
5. Mindfulness (attention control)

“How would you like ChatGPT to respond?”

**Response Style for Learning Support:**

1. **Interactive Approach**
   - Don't present answers immediately; guide with questions (Socratic)
   - Ask "What do you think?" "Why do you think so?"
   - Confirm my understanding before proceeding

2. **Present Scientific Evidence**
   - Show research papers or best practices as basis for suggestions
   - Use careful expressions like "there is a possibility that..."
   - Mention individual differences and limitations

3. **When Proposing Schedules**
   - Consider research findings (lighting color temperature, break timing, etc.)
   - Present specific times and activities
   - Explain why that activity at that time, with scientific basis

4. **When Explaining Code**
   - First confirm my understanding
   - "What do you think this code does?"
   - Identify parts I don't understand before explaining

5. **When Supporting Reflection**
   - First confirm what went well (Keep)
   - Analyze problems (Problem) non-judgmentally
   - Propose specific things to try next (Try)

**Things to Avoid:**
- Presenting answers immediately
- Overly long explanations (keep concise)
- Technical prerequisite explanations (I know as an engineer)
- Excessive encouragement or emotional expressions

Claude Projects Example

Create a dedicated learning assistant in Claude Projects:

Project Name: “Rust Learning Assistant”

Custom Instructions:

You are a Rust learning assistant for an IT engineer who practices scientifically-based learning methods.

**Roles:**
1. Teach Rust concepts interactively (Socratic dialogue)
2. Propose learning schedules based on scientific evidence
3. Support leveraging bugs and prediction violations as learning opportunities
4. Facilitate weekly reflections

**Dialogue Principles:**
- First confirm learner's prior knowledge and predictions
- Don't give answers immediately; guide with questions
- Repeat "Why?" five times to dig deeper
- Support until learner can explain in their own words

**Research Findings to Consider When Proposing Schedules:**
1. Lighting: 6500K for high-difficulty tasks, 5000K for standard, 3500K for light work
2. Breaks: 15min learning + 5min break (AI Learning Pomodoro)
3. Review: Review the morning after learning (24-hour cycle)
4. Sleep: Review before bed, use eye mask
5. Reading: 12 hours of reading per week

**During Code Review:**
- First ask "What do you think this code does?"
- Treat differences between prediction and actual behavior as "surprise learning opportunities"
- Extract learning points from error messages

**During Reflection:**
- Use KPT (Keep, Problem, Try) framework
- Record numerical data (learning time, break count, etc.)
- Make specific improvement suggestions for next week

Project Knowledge (Recommended Uploads):

• Markdown file of previous article “Evidence-Based Effective Learning Methods”
• Markdown file of previous article “Effective Learning Methods Using AI Tools”
• Your own learning logs (exported from Notion/Obsidian)

---

Having AI Create Your Schedule: Practical Prompt Collection

Prompt 1: Create Today’s Optimal Schedule

Please create a scientifically-based learning schedule for today with the following conditions:

**My Situation:**
- Occupation: Full-time engineer (remote work)
- Available learning time: 7:00-8:30 morning, 19:00-21:00 evening
- Learning goal: Understand Rust's ownership system
- Today's tasks: Read "The Rust Book" Ch.4, implement simple code
- Energy level: High in morning, moderate in evening
- AI tools used: Claude, GitHub Copilot

**Research Findings to Consider (from previous articles):**
1. AI Learning Pomodoro (15min learning + 5min break)
2. Lighting color temperature (high difficulty: 6500K, standard: 5000K)
3. 70/30 Rule (70% self, 30% AI)
4. Interactive learning (Socratic dialogue)
5. Pre-sleep review (review next morning)

**Schedule Format:**
Please specify the following for each activity:
- Time and activity content
- Why that activity at that time (scientific basis)
- Lighting setting (color temperature)
- AI usage method
- How to take breaks

Prompt 2: Create Weekly Learning Plan

Please help me create a learning plan for this week:

**My Situation:**
- Occupation: 3rd year university student (Information Engineering major)
- Learning goal: Master Next.js (App Router) and create portfolio site
- Available learning time:
  - Weekdays: 13:00-15:00 (library), 18:30-20:30 (home)
  - Weekends: Saturday 9:00-12:00, 14:00-17:00
- Classes: Monday/Wednesday/Friday mornings, Tuesday/Thursday afternoons
- AI tools used: ChatGPT, Cursor

**This Week's Tasks:**
1. Read Next.js official documentation (Server Components, Routing)
2. Implement simple blog system
3. Deploy (Vercel)

**Research Findings to Emphasize:**
1. 12 hours of reading per week
2. AI Learning Pomodoro
3. Interactive learning
4. Creative activities (4-6 hours per week)

**Plan Format:**
- Day-by-day schedule
- Time allocation for each task
- Scientific basis noted
- Break timing
- Weekend concentrated learning sessions

Prompt 3: Time Blocking Optimization

Please optimize today's schedule based on scientific evidence:

**Current Schedule (tentative):**
- 9:00-12:00 Work (coding)
- 12:00-13:00 Lunch break
- 13:00-18:00 Work (many meetings)
- 19:00-21:00 Rust learning

**Points to Optimize:**
1. Break timing (based on AI Learning Pomodoro)
2. Lighting color temperature change timing
3. Placement of tasks requiring concentration
4. AI usage timing
5. Incorporating creative activities

**Research Findings to Consider:**
- Breaks and memory consolidation
- Lighting color temperature
- Sensory stimulation (music, coffee)
- Mindfulness

**Output Format:**
Please explain the optimized schedule and scientific basis for each change.

Prompt 4: Energy Level-Based Schedule

Please create a learning schedule matched to my energy levels:

**My Energy Pattern:**
- 6:00-9:00: Very high (morning person)
- 9:00-12:00: High
- 12:00-14:00: Low (after lunch)
- 14:00-16:00: Moderate
- 16:00-18:00: Declining
- 18:00-20:00: Moderate (recovering)
- 20:00-22:00: Low

**Learning Tasks (by priority):**
1. Learn difficult Rust concepts (high energy required)
2. Code implementation (moderate energy)
3. Technical book reading (low-moderate energy)
4. Documentation creation (low energy)

**Question:**
Which tasks should be placed at which times?
Please propose with scientific basis (lighting, breaks, etc.).

Prompt 5: Break Optimization

Please optimize breaks in today's learning session:

**Learning Session:**
- Time: 19:00-21:00 (2 hours)
- Task: Implement mini project in Rust
- Tools used: VSCode + GitHub Copilot

**Current Break Plan:**
- 10-minute break at 20:00 (planned)

**Questions:**
1. What's a more effective break timing?
2. What should I do during breaks? (Look at phone vs close eyes)
3. Should I apply AI Learning Pomodoro (15min + 5min)?

Please propose based on scientific evidence (Buch et al. 2021 study, etc.).

---

Persona 1: Full-Time IT Engineer (Remote Work)

Profile

• Name: Tanaka Taro (pseudonym)
• Occupation: Backend engineer (Python/FastAPI)
• Work Style: 5 days/week remote work
• Learning Goal: Master Rust and introduce it to performance-critical parts
• AI Tools Used: GitHub Copilot, Claude, ChatGPT

A Weekday (Typical Example)

6:30  Wake up, breakfast
6:45  Mindfulness meditation (10 min)
      - Breathing meditation with Headspace app
      - Clarify what to learn today

7:00  Environment setup
      - Set smart lights to 6500K (cool white)
      - Place bright blue accents around desk
      - Brew coffee

7:10  Rust Learning Session 1 (AI Learning Pomodoro)
      - 15min: Dialogue with Claude - Review ownership system
        "Let me explain what I learned yesterday in my own words: [explanation]. Is this correct?"
      - 5min break: Close eyes and rehearse what was learned (don't look at phone)

7:30  Implementation Phase
      - 10min: Implement simple Rust code (no AI)
      - 5min break: Stretch, drink water

7:45  Rust Learning Session 2
      - 15min: Dialogue with ChatGPT - Deep dive into borrowing
        Practice "3-Step Thinking Process":
        1. Write prior knowledge in Notion (no AI)
        2. Form predictions and hypotheses
        3. Ask ChatGPT to verify
      - 5min break

8:05  Reflection and Recording
      - Record what was learned today in Notion
      - Record prediction violations in "Surprise Notes"
      - Specify content to review tomorrow morning

9:00  Start work
      - Zoom morning meeting (self-view off)
      - Adjust lighting to 5000K (neutral)

9:30  Coding work
      - Use GitHub Copilot
      - Mindful code review:
        Don't blindly accept Copilot suggestions, understand line by line
        Think "Why this implementation?"

10:30 Short break (5 min)
      - Look out window, light stretch

11:00 Pair programming (Zoom)
      - Turn off self-view function
      - Multi-display layout:
        [Display 1] VSCode + Copilot
        [Display 2] Claude, documentation, Zoom (small)

12:00 Lunch

12:30 Walk (20 min)
      - Walk outside, get natural light
      - Mentally rehearse morning work

12:50 Reading (30 min)
      - Technical book: Read "Programming Rust"
      - Adjust lighting to 3500K (warm white)
      - Put sticky notes on parts not understood

13:30 Resume work (return lighting to 5000K)

15:00 Afternoon break (10 min)
      - Coffee break (before 4pm)
      - Chat with colleagues

18:00 End work
      - Record today's tasks in Notion

18:30 Dinner

19:00 Rust Implementation Project
      - Adjust lighting to 6500K (cool white)
      - Music: Play Lo-fi Hip Hop

19:00 Implementation (25 min)
      - Self-implement (no AI)
      - If bugs occur, "Bug-Driven Learning":
        1. Prediction: "This code should do XX"
        2. Reality: Bug occurs
        3. Record surprise: "Why different from prediction"
        4. Deep dialogue with Claude

19:25 Break (5 min) - Close eyes, brain replay time

19:30 AI Dialogue (25 min)
      - Review implementation with GitHub Copilot Chat
      - "Can this code be improved?"
      - Deep dive "Why is this implementation recommended?"

19:55 Break (5 min)

20:00 Reflection and Recording (15 min)
      - Enter "Learning Log" in Notion
      - Specify content to review tomorrow morning

20:30 Creative Activity (30 min)
      - Create generative art with p5.js
      - Adjust lighting to 3500K (warm white)

21:00 Relaxation time (read novel, talk with family)

21:30 Pre-sleep Review (5-10 min)
      - Recall today's Rust topics (no AI)
      - Try to explain in own words
      - Note parts can't remember (review tomorrow morning)

22:00 Prepare for bed
      - Prepare eye mask, check blackout curtains
      - Adjust room temperature to 18-22°C

22:30 Sleep (with eye mask)

---

[Next morning (6:30)]
6:45  Review yesterday's topics (15 min)
      - Recall what was learned yesterday without AI
      - Check notes recorded in Notion
      - Re-learn parts couldn't remember with Claude/ChatGPT

7:00  Start learning new topic

How to Start the Day with AI

Tanaka-san throws the following prompt to ChatGPT every morning to confirm the day’s optimal schedule:

Morning Schedule Confirmation Prompt:

Good morning. Let me confirm today's learning schedule.

**Yesterday's Learning:**
- Learned basics of Rust's ownership system
- Implemented simple examples
- Got stuck on borrowing

**Today's Situation:**
- Available learning time: 7:10-8:05, 19:00-20:30
- Today's tasks: Understand borrowing, introduce lifetimes
- Energy level: High in morning, moderate in evening
- Work: 9:00-12:00 coding, 13:30-18:00 many meetings

**Questions:**
1. Should today's learning schedule be same as yesterday's plan? Any adjustments needed?
2. Is learning both borrowing and lifetimes on the same day too much?
3. How much time should I allocate to borrowing (what I got stuck on yesterday) in the morning session?

Please propose with scientific basis.

Implementation Methods

1. Morning Mindfulness Meditation

• Use apps like Headspace, Calm, or Waking Up
• Start day with 10-minute breathing meditation
• Clarify what to learn today

2. Thorough AI Learning Pomodoro

• Strictly follow 15min learning + 5min break cycle
• Use timer apps (Pomofocus, Toggl, etc.)
• During breaks, don’t look at phone; close eyes to secure brain replay time

3. Lighting Optimization

• Adjust color temperature with smart lights (Philips Hue, LIFX, etc.)
• High-difficulty tasks (Rust learning): 6500K
• Standard work: 5000K
• Reading/relaxation: 3500K

4. How to Dialogue with AI

• Practice Socratic dialogue
• Confirm “My understanding is XX. Is this correct?”
• Repeat “Why?” five times to dig deeper

5. Bug-Driven Learning

• View bugs as “learning opportunities” not “problems”
• Record differences between prediction and actual behavior
• Dialogue with Claude to understand root causes

6. 24-Hour Cycle Review

• Learning at night → Review before sleep → Re-review next morning
• Leverage memory consolidation during sleep

Scientific Basis

Mindfulness ¹: Dumontheil et al. (2023) research shows that 8 weeks of mindfulness training improves attention reorientation. Morning meditation helps with attention control when using AI.

Lighting Color Temperature ²: Choi & Suk (2016) research shows optimal color temperatures:

• 6500K: High-difficulty tasks requiring concentration
• 5000K: Standard learning activities
• 3500K: Relaxation activities, reading

Bright Colors ³: Al-Ayash et al. (2016) research shows significantly improved reading scores in environments with bright colors (especially blue and yellow).

Breaks and Memory Consolidation ⁴: Buch et al. (2021) research shows that during breaks, the brain replays learning content (~20x time compression), promoting memory consolidation. Using phone during breaks may interfere with this replay.

Interactive Learning ⁵: Zavala & Kuhn (2017) research shows that interactive approaches (Socratic dialogue) promote deeper, more comprehensive processing.

Bug-Driven Learning (Leveraging Surprise) ⁶: Frank, Kafkas, & Montaldi (2022) research shows that surprise (prediction violations) strengthens memory encoding.

Self-View Off ⁷: Tien, Imundo, & Bjork (2023) research shows that viewing one’s own image during online learning increases appearance anxiety and decreases memory performance.

Creative Activities ⁸: Ishiguro, Ishihara, & Morita (2023) research shows that extracurricular activities in music and visual arts are associated with improved academic performance.

Sleep Quality ⁹: Greco et al. (2023) research shows that wearing an eye mask improves sleep quality and episodic memory. The 24-hour cycle (learning → sleep → morning review) promotes memory consolidation.

Caveats

• Individual Differences: The same schedule doesn’t suit everyone. Adjust to your own energy patterns.
• Flexibility: Some days won’t go as planned. Don’t aim for perfection; target 80% execution.
• Over-Planning: Don’t spend too much time on planning; focus on execution.
• AI Dependence: Follow the 70/30 rule; don’t rely too much on AI.

---

Persona 2: University Student / Programming Learner

Profile

• Name: Sato Hanako (pseudonym)
• Occupation: 3rd year university student (Information Engineering major)
• Learning Goal: Master web development (React, Next.js) and apply for internships
• AI Tools Used: ChatGPT, Cursor, Claude

A Weekday (Class Day)

6:30  Wake up, breakfast

7:00  Morning Learning Session (30 min)
      - Lighting: 6500K (cool white)
      - Review yesterday's Next.js topic (no AI)
      - Check parts couldn't remember with ChatGPT

7:30  Commute to university (read tech blogs during commute)

8:30  Class 1: Algorithms and Data Structures
      - Handwrite lecture notes (paper)
      - Put sticky notes on questions

10:30 Break (10 min) - Chat with friends, light movement

10:40 Class 2: Database Design

12:30 Lunch (30 min) - Cafeteria with friends

13:00 Study at Library (2 hours)
      - Environment: Quiet, natural light

13:00 Algorithm Assignment (AI Learning Pomodoro)
      - 15min: Solve problem yourself (no AI)
        Think for yourself first, write algorithm on paper
      - 5min break: Close eyes, organize thoughts
      - 15min: Dialogue with ChatGPT
        "My solution is XX. Is this correct?"
        "Is there a more efficient method?"
      - 5min break
      - 10min: Implement optimal solution (using Cursor)
        Understand each line before committing
      - 5min break

13:55 Database Assignment (30 min) - Same cycle

14:30 Technical Book Reading (30 min)
      - Read "Learning React"
      - Put sticky notes on unclear parts

15:00 Class 3: Software Engineering

17:00 Return home (read tech blogs on train)

18:00 Dinner

18:30 Next.js Learning Session (2 hours)
      - Lighting: 6500K (cool white)
      - Desk area: Bright blue notebooks, stationery
      - Play Lo-fi Hip Hop

18:30 Read Official Documentation (20 min)
      - About Server Components
      - "3-Step Thinking Process":
        1. Prior knowledge: Write "How is this different from SSR I know?"
        2. Prediction: Hypothesize "It probably does XX"
        3. Read documentation to verify

18:50 Break (5 min)

18:55 Dialogue with Claude (20 min)
      - "My understanding of Server Components is XX. Is this correct?"
      - Deep dive with Socratic dialogue

19:15 Break (5 min)

19:20 Implementation (30 min)
      - Create simple Next.js app with Cursor
      - Design yourself first (no AI)
      - Use Cursor for implementation

19:50 Break (5 min)

19:55 Bug Fixing and Debugging (25 min)
      - If bugs occur, treat as "surprise learning opportunity"
      - Dialogue with Claude to understand root cause

20:20 Reflection (10 min)
      - Enter "Learning Log" in Notion

20:30 Creative Activity (30 min)
      - Portfolio site design in Figma
      - Or hobby instrument playing (guitar)
      - Lighting: 3500K (warm white)

21:00 Reading (30 min) - Novel or technical book

21:30 Pre-sleep Review (10 min)
      - Recall today's Next.js content (no AI)
      - Note parts couldn't remember

22:00 Prepare for bed (prepare eye mask, check blackout curtains)

22:30 Sleep

How to Create Weekly Plans with AI

Sato-san throws the following prompt to ChatGPT every Sunday night to plan the following week:

Weekly Planning Prompt:

Please help me create a learning plan for next week.

**Last Week's Reflection:**
- Total learning time: 15 hours
- What went well:
  - Concentrated learning at library (quiet environment)
  - AI Learning Pomodoro practice
- Needs improvement:
  - Weekend study time was short (Saturday only 2 hours)
  - Reading time didn't reach goal (12 hours/week) (actual 8 hours)

**Next Week's Situation:**
- Classes: Monday/Wednesday/Friday mornings, Tuesday/Thursday afternoons
- Learning goal: Understand Next.js routing and data fetching
- Available learning time:
  - Weekdays: 13:00-15:00 (library), 18:30-20:30 (home)
  - Weekends: Saturday 9:00-12:00, 14:00-17:00, Sunday 9:00-12:00
- Want to do:
  - Next.js implementation
  - Solve LeetCode problems (3-4/week)
  - Reading (want to achieve 12 hours/week)
  - Creative activity (design in Figma)

**Questions:**
1. How should I incorporate 12 hours of reading per week?
2. What's the balance between LeetCode and Next.js learning?
3. Tips to increase weekend learning time?

Please propose based on scientific evidence (research from previous articles).

Implementation Methods

1. Utilize Library Learning

• Concentration improves in quiet environment
• Choose seats with natural light
• Practice AI Learning Pomodoro (15min + 5min)

2. Effective Use of Commute Time

• Read tech blogs and articles on train
• Count toward 12 hours/week reading habit
• Use reading apps (Pocket, Instapaper, etc.) on smartphone

3. Handwriting Lecture Notes

• Record in paper notebook by hand
• More memorable than digital
• Put sticky notes on questions, ask AI later

4. Implementation with Cursor

• First think through design yourself (no AI)
• Use Cursor’s AI completion during implementation
• Understand suggested code before adopting

5. Weekend Concentrated Learning

• Difficult topics during high-energy morning hours
• Afternoon for implementation and projects
• Always incorporate creative activities

Scientific Basis

12 Hours of Reading per Week ¹⁰: Sun et al. (2023) research shows that 12 hours of reading per week has positive effects on cognitive ability and mental health. Aim for 12 hours including technical books, blogs, and novels.

Other basis refers to same research as Persona 1

Caveats

• Balancing Classes and Learning: Don’t overdo it; prioritize class assignments.
• Ensure Sleep Time: Don’t cut sleep to extend learning time.
• Time with Friends: Value time with friends, not just studying.

---

Persona 3: Freelance Engineer

Profile

• Name: Yamada Jiro (pseudonym)
• Occupation: Freelance full-stack engineer
• Work Style: 3-4 days/week client work, rest self-learning/side projects
• Learning Goal: Master machine learning (PyTorch) and acquire ML projects
• AI Tools Used: Claude, ChatGPT, GitHub Copilot

A Learning Day (Friday - No Client Work)

7:00  Wake up, breakfast

7:30  Mindfulness Meditation (15 min)
      - Reflect on this week's learning
      - Clarify today's learning goals

7:45  Morning Walk (30 min)
      - Walk in neighborhood, mentally rehearse week's learnings

8:15  Environment Setup
      - Lighting: 6500K (cool white)
      - Desk: Bright color accents
      - Brew coffee
      - Play classical music (Bach)

8:30  PyTorch Learning Session 1 (2 hours - AI Learning Pomodoro)

8:30  Book Reading (25 min)
      - "Deep Learning with PyTorch" Chapter 3
      - Put sticky notes on unclear parts

8:55  Break (5 min) - Close eyes, rehearse content

9:00  Dialogue with Claude (25 min)
      - "My understanding of tensor operations is XX. Is this correct?"
      - Deep dive with Socratic dialogue

9:25  Break (5 min)

9:30  Implementation (25 min)
      - Implement simple neural network (no AI)
      - If bugs occur, treat as "surprise learning opportunity"

9:55  Break (5 min)

10:00 Collaborate with GitHub Copilot (25 min)
      - See Copilot suggestions
      - Ask Copilot Chat "Why this implementation?"

10:25 Break (10 min) - Stretch, coffee break

10:35 Reflection and Recording (15 min)
      - Enter "Learning Log" in Obsidian

10:50 Machine Learning Project (3 hours)
      - Build image classification model with real dataset
      - Lighting: Maintain 6500K
      - Music: Electronic (maintain concentration)

10:50 Data Preprocessing (50 min - no AI)

11:40 Break (10 min)

11:50 Model Building (50 min - collaborate with GitHub Copilot)

12:40 Break (10 min)

12:50 Debugging and Tuning (50 min)
      - Dialogue with Claude to deep dive causes

13:40 Lunch (1 hour 20 min)

15:00 Technical Article Writing (2 hours)
      - Lighting: 5000K (neutral)
      - Blog article about this week's PyTorch learnings

15:00 Think through structure (30 min - no AI)

15:30 Writing (60 min)
      - Ask Claude "Please review"

16:30 Break (10 min)

16:40 Finish article (20 min)

17:00 Break / light snack (30 min)

17:30 Music Production (1 hour 30 min)
      - Song creation with DAW (Ableton Live)
      - Lighting: 3500K (warm white)
      - Activate different brain regions

19:00 Dinner (1 hour)

20:00 Technical/Business Book Reading (2 hours)
      - Lighting: 3500K (warm white)

20:00 "Designing Data-Intensive Applications" (1 hour)

21:00 Novel or general book (1 hour)

22:00 Pre-sleep Review (15 min)
      - Recall today's PyTorch content (no AI)
      - Record in Obsidian

22:15 Prepare for bed

22:45 Sleep

Monthly Goal Setting with AI

Yamada-san throws the following prompt to Claude at the beginning of each month to set monthly goals and milestones:

Monthly Goal Setting Prompt:

Please help me set learning goals and milestones for this month.

**Last Month's Reflection:**
- Goal: Master PyTorch basics
- Achievement rate: 80% (understood tensor operations, autograd, basic NN construction)
- Not achieved: Only surface understanding of CNN and RNN
- Total learning time: 60 hours (15 hours/week)
- Reading time: 45 hours (11 hours/week - slightly short of goal)

**This Month's Big Goals:**
1. Deeply understand and implement CNN/RNN in PyTorch
2. Participate in beginner Kaggle competition
3. Acquire 1 ML-related project

**Constraints:**
- Client work: 3 days/week (Mon/Tue/Wed)
- Available learning days: 2 days/week (Thu/Fri) + weekends
- Target learning time: 16 hours/week

**Questions:**
1. Is this goal realistic? Any adjustments needed?
2. How should I allocate 16 hours/week to CNN learning, RNN learning, Kaggle, and project acquisition?
3. Please propose weekly milestones
4. How should I incorporate 12 hours of reading per week?

Please propose considering scientific evidence and my past learning pace.

Implementation Methods

1. Long Concentrated Learning Sessions

• Divide 2-3 hour concentration sessions with AI Learning Pomodoro
• Repeat 25min + 5min cycles
• Include longer 10min breaks in between

2. Technical Article Writing

• Output what was learned
• Ask Claude for review
• Use as opportunity to confirm own understanding

3. Prioritize Creative Activities

• Music production, 3D modeling, design, etc.
• Stimulate different brain regions
• Secure 4-6 hours/week

4. Ensure Reading Time

• Technical books: 6-7 hours/week
• General books/novels: 5-6 hours/week
• Target total 12 hours/week

Scientific Basis

Refers to same research as Persona 1.

Caveats

• Balance with Client Work: Don’t lower client work quality by focusing too much on learning.
• Secure Income: Continue learning while maintaining income sources as freelancer.
• Prevent Isolation: Working alone for long periods; maintain regular communication with peers.

---

Weekly Review with AI

Conduct weekly review with AI every Sunday night.

Weekly Review Prompt

Please reflect on this week's learning together and think about improvements for next week.

**This Week's Record:**

**Learning Time:**
- Total learning time: 18 hours
- AI usage time: 6 hours
- Non-AI (self) time: 12 hours (70/30 rule achieved!)
- Reading time: 14 hours (goal 12 hours achieved!)

**Research Findings Practice Status:**

1. Lighting Optimization (Research 2):
   - Days practiced: 5/7 days
   - Felt effect: 4/5 (concentration improved)
   - Notes: Couldn't practice Wed and Sat due to going out

2. AI Learning Pomodoro (Research 9):
   - Days practiced: 6/7 days
   - Felt effect: 5/5 (realized importance of breaks)
   - Notes: Couldn't practice Sunday due to long reading session

3. Interactive Learning (Research 5):
   - Days practiced: 7/7 days
   - Felt effect: 5/5 (deepened understanding)
   - Notes: Socratic dialogue with Claude especially effective

4. 24-Hour Cycle Review (Research 12):
   - Days practiced: 4/7 days
   - Felt effect: 3/5 (not yet habit)
   - Notes: Tend to forget morning review

5. Creative Activities (Research 11):
   - Weekly activity time: 5 hours (goal 4-6 hours achieved)
   - Activity content: Generative art with p5.js, guitar practice
   - Felt effect: 4/5 (refreshed)

**What Went Well (Keep):**
- Concentration improved with AI Learning Pomodoro practice
- Reading time achieved goal (14 hours/week)
- Understanding deepened with Socratic dialogue with Claude

**Needs Improvement (Problem):**
- 24-hour cycle review not becoming habit (forget morning review)
- Weekend learning time short (only Sat 3 hours, Sun 2 hours)
- Sleep time irregular (targeting 23:00-7:00, actually 24:00-8:00 often)

**Try Next Week:**
- Set "yesterday's review" reminder in morning alarm
- Increase weekend learning time (target Sat 6 hours, Sun 4 hours)
- Set bedtime to 22:30 (gradually earlier)

**Questions:**
1. Tips to make 24-hour cycle review a habit?
2. How to increase weekend learning time? (No motivation)
3. Specific methods to move bedtime earlier?
4. What should be adjusted in next week's learning plan?

Please advise considering scientific evidence and my past patterns.

AI Response Example (Claude)

Great results from this week's learning! Especially great achieving reading time and creative activity goals. Let's analyze the improvements in detail.

---

## Tips to Make 24-Hour Cycle Review a Habit

**Problem Analysis:**
"Forgetting" morning review means it's not yet a habit. Habituation requires triggers and routines.

**Specific Suggestions:**

1. **Set Triggers**
   - Brewing morning coffee → Open Notion "yesterday's review notes" meanwhile
   - Right after mindfulness meditation → 5-minute review time
   - After breakfast → Secure 5-10 minutes as "review time"

2. **Simplify Review**
   - Shorten review time to 5 minutes (currently 10min → 5min)
   - Simple format:
     - "3 things learned yesterday"
     - "1 thing couldn't remember"
     - "What to confirm in today's morning session"

3. **Improve Notion Template**
   - During pre-sleep review, "clearly" describe next morning's review content
   - Current: "Review borrowing" (vague)
   - Improved: "Can I explain 3 rules of borrowing in my own words?" (specific)

**Scientific Basis:**
The "trigger → routine → reward" loop is important for habituation. Use morning coffee as trigger, review as routine, and "confirmation of understanding" as reward.

---

## How to Increase Weekend Learning Time

**Problem Analysis:**
"No motivation" may be due to:
1. Accumulated fatigue from weekdays
2. Vague weekend tasks
3. Hard to get reward (sense of achievement)

**Specific Suggestions:**

1. **Plan Weekend on Friday Night**
   - Dialogue with Claude at 21:00 Friday
   - Throw "Create tomorrow (Saturday) learning plan" prompt
   - Specific tasks ("Read Ch.5 of Rust" → "Understand Rust structs and implement 3 examples")

2. **Start Morning Earlier**
   - Target: Wake 8:00 Sat → Start learning 9:00
   - 1 hour earlier than current wake time (probably 9:00-10:00)

3. **Set Rewards**
   - After 3 hours morning learning, eat favorite lunch
   - After 3 hours afternoon learning, watch movie, meet friends, etc.

4. **Change Environment**
   - Saturday morning: Study at cafe (change environment)
   - Saturday afternoon: Implementation at home
   - Sunday: Concentrated learning at library

**Scientific Basis:**
Environmental change brings freshness and increases motivation. Also, setting specific tasks and rewards increases sense of achievement and promotes continuous learning.

---

## How to Move Bedtime Earlier

**Problem Analysis:**
Changing from 24:00 to 22:30 bedtime requires 1.5 hour adjustment. Don't change all at once; adjust gradually.

**Specific Suggestions (4-Week Plan):**

**Week 1 (Next Week): Target 23:30 bedtime**
- Start bedtime preparation at 23:00
- Change lighting to 3500K (warm white) from 22:30
- No phone/PC after 22:30
- 23:00-23:30: Reading (novel, relaxation)

**Week 2: Target 23:15 bedtime**
- 15 minutes earlier

**Week 3: Target 23:00 bedtime**

**Week 4: Target 22:30 bedtime**

**Support Measures:**
- Use eye mask (Research 12: Greco et al. 2023)
- Adjust room temperature to 18-22°C
- Avoid caffeine after afternoon (non-caffeine after 16:00)
- Shower 1 hour before bed (lower body temp to promote sleepiness)

**Scientific Basis:**
Greco et al. (2023) research shows sleep quality affects learning. Gradually moving bedtime earlier allows habit change without strain.

---

## Next Week's Learning Plan Adjustment Suggestions

**Adjustment 1: Make Morning Review a Habit**
- Incorporate "5-minute review" into morning routine
- Improve Notion template

**Adjustment 2: Increase Weekend Learning Time**
- Plan weekend at 21:00 Friday
- Target Sat 6 hours, Sun 4 hours (current: Sat 3 hours, Sun 2 hours)
- Set specific tasks

**Adjustment 3: Adjust Bedtime**
- Target 23:30 bedtime next week
- Reduce screen time after 22:30

**Adjustment 4: Continue AI Learning Pomodoro**
- Current 6/7 days practice → Target 7/7 days
- Include 10-minute break every 2 hours even during Sunday reading time

---

**Next Week's Target Score Prediction:**
- Total learning time: 18 hours → 22 hours (weekend +4 hours)
- 24-hour cycle review: 4/7 days → 6/7 days
- Bedtime: 24:00 → 23:30

Let's improve gradually. Don't aim for perfection; 80% execution is enough!

Implementation Methods

1. Fix Weekly Review Time

• Conduct every Sunday 21:00-21:30
• Block in calendar

2. Record Numerical Data

• Auto-record learning time with Toggl, RescueTime, etc.
• Checklist research findings practice status in Notion

3. Use KPT Framework

• Keep (continue doing)
• Problem (needs improvement)
• Try (try next week)

4. Dialogue with AI

• Not just recording; dialogue with AI to get improvement suggestions
• Request advice based on scientific evidence

Scientific Basis

Weekly review enhances metacognition (ability to overview one’s own learning). Wade & Kidd (2019) research¹¹ shows that recognizing one’s own knowledge state is important for learning.

Caveats

• Avoid Perfectionism: Don’t aim for 100% execution; 80% is sufficient.
• Flexibility: Some weeks won’t go as planned. Don’t blame yourself; apply to next week.
• Long-term Perspective: Changes in 1 week are small, but continuing 8-12 weeks produces big changes.

---

Summary: Practicing Evidence-Based AI Learning

This article demonstrated how to practically integrate the 14 research findings introduced in the two previous articles through a day in the life of 3 different personas (full-time engineer, student, freelancer).

Additionally, specific prompt examples for optimizing learning with AI tools were provided.

Common Important Points

1. Setting Up AI Personal Assistant

Create a personal learning assistant with ChatGPT Custom Instructions or Claude Projects. By incorporating scientifically-based learning methods, AI can propose optimal schedules and learning methods.

2. Having AI Create Your Schedule

With prompts like “Create today’s optimal schedule,” you can get schedules proposed with research findings in mind. Save time thinking while learning based on scientific evidence.

3. Thorough AI Learning Pomodoro⁴

All personas adopt the 15min learning + 5min break cycle. During breaks, don’t look at phone; secure brain replay time by closing eyes.

4. 70/30 Rule¹²

Spend 70% of learning time thinking yourself, 30% relying on AI. Don’t depend too much on AI; maintain active learning.

5. 24-Hour Cycle Review⁹

Learning at night → Sleep → Morning review cycle promotes memory consolidation and retention.

6. Use AI in Weekly Review

Conduct reflection with AI every Sunday, have it propose improvements for next week. Use KPT framework (Keep, Problem, Try).

3 Actions to Start Today

Level 1: Start Now (5 min)

1. Set ChatGPT Custom Instructions (copy-paste above example, customize for yourself)
2. Throw “Create today’s optimal schedule” prompt to AI
3. Set 15min + 5min break with timer app

Level 2: Start Today (30 min)

1. Create “Learning Log” template in Notion/Obsidian
2. Create dedicated learning assistant in Claude Projects
3. Save weekly review prompt

Level 3: Start This Week (2 hours)

1. Environment setup (introduce smart lights, purchase eye mask)
2. Customize schedule of persona closest to your situation
3. Conduct weekly review (block Sunday 21:00 in schedule)

Finally: AI as Learning Partner

AI tools are not just “tools that give answers,” but learning partners.

Using the prompts introduced in this article, AI serves these roles:

• Schedule Proposer: Proposes optimal schedules based on scientific evidence
• Dialogue Partner: Deepens understanding through Socratic dialogue
• Reviewer: Verifies code and understanding
• Coach: Proposes improvements in weekly reflection

Combined with the two previous articles, I hope this article helps improve your learning habits.

Combine scientifically evidence-based learning methods with the power of AI tools to achieve continuous growth.

---

References

The research cited in this article is explained in detail in the following articles:

Related Articles

• Evidence-Based Effective Learning Methods: Optimizing Environment, Methodology, and Habits - Detailed explanation of 14 research papers
• Effective Learning Methods Using AI Tools: Evidence-Based Practical Guide - Application methods for AI tools

Citation Numbers and Corresponding Research

The citation numbers in this article correspond to the same 14 studies from the previous articles:

For detailed reference list (author information, sample sizes, research limitations, etc.), please refer to the related articles above.

1. Dumontheil et al. (2023) - Mindfulness - Journal of Adolescence. [Reliability: High] ↩︎

2. Choi & Suk (2016) - Lighting color temperature - Optics Express. [Reliability: High] ↩︎

3. Al-Ayash et al. (2016) - Color influence - Color Research & Application. [Reliability: High] ↩︎

4. Buch et al. (2021) - Breaks and memory consolidation - Cell Reports. [Reliability: High] ↩︎ ↩︎²

5. Zavala & Kuhn (2017) - Interactive learning - Psychological Science. [Reliability: High] ↩︎

6. Frank, Kafkas, & Montaldi (2022) - Surprise and memory - Journal of Neuroscience. [Reliability: High] ↩︎

7. Tien, Imundo, & Bjork (2023) - Online learning and self-view (same as ¹³) - Applied Cognitive Psychology. [Reliability: High] ↩︎

8. Ishiguro, Ishihara, & Morita (2023) - Creative activities - npj Science of Learning. [Reliability: High] ↩︎

9. Greco et al. (2023) - Sleep quality - Sleep. [Reliability: High] (Note: Some results questioned by reanalysis paper in August 2023) ↩︎ ↩︎²

10. Sun et al. (2023) - Reading habits - Psychological Medicine. [Reliability: High] ↩︎

11. Wade & Kidd (2019) - Curiosity and prior knowledge - Psychonomic Bulletin & Review. [Reliability: High] ↩︎

12. Buchan et al. (2020) - Active learning - npj Science of Learning. [Reliability: High] ↩︎

13. Tien, Imundo, & Bjork (2023) - Online learning and self-view - Applied Cognitive Psychology. [Reliability: High] ↩︎