사람과 AI의 문장 판단 차이 분석하기
Students change one design variable, run a safe task, and compare what happened.
Task result, time or errors, and user feedbackCOLLEGE APPLICATION RESEARCH EVIDENCE · 고등 · GRADES 9-12
Human-AI Research LabHuman Experience Engineering & Affective AI
How can we measure and improve experience without oversimplifying people or enabling manipulation?
Students build a study or system that shows uncertainty, checks bias, protects user choice, and explains its limits. Choose online or in-person classes in Apgujeong, in a one-to-one or small-group format.
The course was designed from open research published by Carnegie Mellon and Stanford. Students first explain how university researchers study whether technology is useful, easy, fair, and safe. They then distinguish their own question, design choice, A/B result, failure, revision, and limit for an authentic college application or interview explanation.
What students build
Build one project
through 12 evidence steps.
- Show the reason for a recommendation and its uncertainty, or abstain.
- Let the user correct, override, opt out, and delete.
- Use only the minimum non-sensitive task data the user chose to share.
Tool path
Figma, surveys, and spreadsheetsPython and JupyterSynthetic or public dataOptional non-sensitive sensors
Portfolio
- Research question and measurement plan
- System map
- Model or system card
- Failure analysis
- Research poster and reproducibility notes
Experience lenses
Emotional experienceThinking experienceSensory and physical experienceAction experienceSocial experienceTime and context
12-session roadmap
Visible work and
one next decision each time.
Open any session to see the mission, outputs, project step, and short family summary.
01감정 기술의 과장된 주장 확인하기감정 기술의 과장된 주장 확인하기
Students complete Emotion-Tech Mythbusters, make one clear design choice, test it, and record what changed.
+
감정 기술의 과장된 주장 확인하기
Students complete Emotion-Tech Mythbusters, make one clear design choice, test it, and record what changed.
Experience lenses
Emotional experienceThinking experienceSocial experience
Outputs
- Design notes
- Prototype or test plan
- Evidence and next change
Project step
Add this session's evidence to the Human-Friendly Adaptive Study Assistant project.
Family summary
Today, students completed Emotion-Tech Mythbusters. They changed one design variable, checked the result, and explained the next improvement.
02보이지 않는 경험을 관찰 가능한 기준으로 정하기보이지 않는 경험을 관찰 가능한 기준으로 정하기
Students complete Measure the Unmeasurable, make one clear design choice, test it, and record what changed.
+
보이지 않는 경험을 관찰 가능한 기준으로 정하기
Students complete Measure the Unmeasurable, make one clear design choice, test it, and record what changed.
Experience lenses
Thinking experienceEmotional experienceAction experienceTime and context
Outputs
- Design notes
- Prototype or test plan
- Evidence and next change
Project step
Add this session's evidence to the Human-Friendly Adaptive Study Assistant project.
Family summary
Today, students completed Measure the Unmeasurable. They changed one design variable, checked the result, and explained the next improvement.
03이해하고 철회할 수 있는 동의 화면 만들기이해하고 철회할 수 있는 동의 화면 만들기
Students complete Consent by Design, make one clear design choice, test it, and record what changed.
+
이해하고 철회할 수 있는 동의 화면 만들기
Students complete Consent by Design, make one clear design choice, test it, and record what changed.
Experience lenses
Time and contextThinking experienceSocial experienceAction experience
Outputs
- Design notes
- Prototype or test plan
- Evidence and next change
Project step
Add this session's evidence to the Human-Friendly Adaptive Study Assistant project.
Family summary
Today, students completed Consent by Design. They changed one design variable, checked the result, and explained the next improvement.
04알림이 많은 화면에서 중요한 신호 찾기알림이 많은 화면에서 중요한 신호 찾기
Students complete Control Room Under Pressure, make one clear design choice, test it, and record what changed.
+
알림이 많은 화면에서 중요한 신호 찾기
Students complete Control Room Under Pressure, make one clear design choice, test it, and record what changed.
Experience lenses
Thinking experienceSensory and physical experienceAction experience
Outputs
- Design notes
- Prototype or test plan
- Evidence and next change
Project step
Add this session's evidence to the Human-Friendly Adaptive Study Assistant project.
Family summary
Today, students completed Control Room Under Pressure. They changed one design variable, checked the result, and explained the next improvement.
05AI 추천이 실패하는 단계와 안전장치 그리기AI 추천이 실패하는 단계와 안전장치 그리기
Students complete Build the Affective Pipeline, make one clear design choice, test it, and record what changed.
+
AI 추천이 실패하는 단계와 안전장치 그리기
Students complete Build the Affective Pipeline, make one clear design choice, test it, and record what changed.
Experience lenses
Thinking experienceAction experienceTime and contextSocial experience
Outputs
- Design notes
- Prototype or test plan
- Evidence and next change
Project step
Add this session's evidence to the Human-Friendly Adaptive Study Assistant project.
Family summary
Today, students completed Build the Affective Pipeline. They changed one design variable, checked the result, and explained the next improvement.
06사람과 AI의 문장 판단 차이 분석하기사람과 AI의 문장 판단 차이 분석하기
Students complete Sentiment Model on Trial, make one clear design choice, test it, and record what changed.
+
사람과 AI의 문장 판단 차이 분석하기
Students complete Sentiment Model on Trial, make one clear design choice, test it, and record what changed.
Experience lenses
Emotional experienceThinking experienceSocial experienceTime and context
Outputs
- Design notes
- Prototype or test plan
- Evidence and next change
Project step
Add this session's evidence to the Human-Friendly Adaptive Study Assistant project.
Family summary
Today, students completed Sentiment Model on Trial. They changed one design variable, checked the result, and explained the next improvement.
07관찰 신호와 사람의 실제 마음 구분하기관찰 신호와 사람의 실제 마음 구분하기
Students complete Signal Is Not Inner State, make one clear design choice, test it, and record what changed.
+
관찰 신호와 사람의 실제 마음 구분하기
Students complete Signal Is Not Inner State, make one clear design choice, test it, and record what changed.
Experience lenses
Emotional experienceSocial experienceTime and contextAction experience
Outputs
- Design notes
- Prototype or test plan
- Evidence and next change
Project step
Add this session's evidence to the Human-Friendly Adaptive Study Assistant project.
Family summary
Today, students completed Signal Is Not Inner State. They changed one design variable, checked the result, and explained the next improvement.
08AI가 모를 때 멈추고 묻게 하는 화면 만들기AI가 모를 때 멈추고 묻게 하는 화면 만들기
Students complete Uncertainty UX Lab, make one clear design choice, test it, and record what changed.
+
AI가 모를 때 멈추고 묻게 하는 화면 만들기
Students complete Uncertainty UX Lab, make one clear design choice, test it, and record what changed.
Experience lenses
Thinking experienceEmotional experienceAction experienceSocial experience
Outputs
- Design notes
- Prototype or test plan
- Evidence and next change
Project step
Add this session's evidence to the Human-Friendly Adaptive Study Assistant project.
Family summary
Today, students completed Uncertainty UX Lab. They changed one design variable, checked the result, and explained the next improvement.
09Research and Architecture ReviewResearch and Architecture Review
Students complete Research and Architecture Review, make one clear design choice, test it, and record what changed.
+
Research and Architecture Review
Students complete Research and Architecture Review, make one clear design choice, test it, and record what changed.
Experience lenses
Thinking experienceAction experienceSocial experienceTime and context
Outputs
- Design notes
- Prototype or test plan
- Evidence and next change
Project step
Add this session's evidence to the Human-Friendly Adaptive Study Assistant project.
Family summary
Today, students completed Research and Architecture Review. They changed one design variable, checked the result, and explained the next improvement.
10사용자가 통제하는 시험용 AI 만들기사용자가 통제하는 시험용 AI 만들기
Students complete Prototype Implementation Sprint, make one clear design choice, test it, and record what changed.
+
사용자가 통제하는 시험용 AI 만들기
Students complete Prototype Implementation Sprint, make one clear design choice, test it, and record what changed.
Experience lenses
Thinking experienceAction experienceSensory and physical experienceTime and context
Outputs
- Design notes
- Prototype or test plan
- Evidence and next change
Project step
Add this session's evidence to the Human-Friendly Adaptive Study Assistant project.
Family summary
Today, students completed Prototype Implementation Sprint. They changed one design variable, checked the result, and explained the next improvement.
11Pilot Study and Failure AnalysisPilot Study and Failure Analysis
Students complete Pilot Study and Failure Analysis, make one clear design choice, test it, and record what changed.
+
Pilot Study and Failure Analysis
Students complete Pilot Study and Failure Analysis, make one clear design choice, test it, and record what changed.
Experience lenses
Action experienceThinking experienceEmotional experienceSocial experienceTime and context
Outputs
- Design notes
- Prototype or test plan
- Evidence and next change
Project step
Add this session's evidence to the Human-Friendly Adaptive Study Assistant project.
Family summary
Today, students completed Pilot Study and Failure Analysis. They changed one design variable, checked the result, and explained the next improvement.
12최종 프로젝트의 결과와 한계 발표하기최종 프로젝트의 결과와 한계 발표하기
Students complete Capstone Defence, make one clear design choice, test it, and record what changed.
+
최종 프로젝트의 결과와 한계 발표하기
Students complete Capstone Defence, make one clear design choice, test it, and record what changed.
Experience lenses
Emotional experienceThinking experienceSensory and physical experienceAction experienceSocial experienceTime and context
Outputs
- Design notes
- Prototype or test plan
- Evidence and next change
Project step
Add this session's evidence to the Human-Friendly Adaptive Study Assistant project.
Family summary
Today, students completed Capstone Defence. They changed one design variable, checked the result, and explained the next improvement.
Highlighted practicals
More investigating,
making, and testing.
Every class has a named activity, one variable to change, evidence to collect, and a low-tech option.
개인정보의 수집·사용·삭제 과정 그리기
Students change one design variable, run a safe task, and compare what happened.
Task result, time or errors, and user feedback알림이 많은 화면에서 중요한 신호 찾기
Students change one design variable, run a safe task, and compare what happened.
Task result, time or errors, and user feedbackAI가 모른다고 알리는 화면 만들기
Students change one design variable, run a safe task, and compare what happened.
Task result, time or errors, and user feedbackAlternative final projects
Same skills.
Different interests.
Students can change the topic while keeping the same research, A/B test, evidence, access, and ethics standard.
High school · Interactive
Kansei Product Experience Optimisation
The link between design choices and trust, comfort, or clarity has not been tested.
See process and evidence →High school · Interactive
감정 자료를 나누는 AI의 편향과 한계 확인
Average performance hides failures in specific contexts and disagreement between people.
See process and evidence →High school · Interactive
User-Controlled Adaptive Game
The game guesses the player's state, changes the level, and does not explain why.
See process and evidence →High school · Interactive
Human-Friendly AI Assistant
The model gives a definite answer even when the context is unclear.
See process and evidence →High school · Interactive
Responsible Notification Engine
Optimising engagement increases interruption and false urgency.
See process and evidence →High school · Physical
Human-Robot Handoff Experience
We do not yet know how distance, speed, timing, and feedback affect safety and trust.
See process and evidence →High school · Interactive
AI Explanation & Trust Calibration Study
A long or overconfident explanation can make trust less accurate.
See process and evidence →High school · Interactive
Multilingual Sentiment Limitations Study
People and models often disagree on sarcasm, indirect language, and mixed-language text.
See process and evidence →Assessment
Grade the research
and engineering, not the person.
We assess the quality of the question, evidence, build, test, improvement, access, and stated limits.
사용자 이해와 증거 수집15%
필요를 설계 요구사항으로 번역15%
시험용 작품·기술 구현25%
테스트·분석·반복 개선30%
윤리·포용·한계·소통15%
Safety boundary
Do not claim to know
a person's inner state.
Use safe tasks, minimal data, clear choices, and visible limits.
- No emotion diagnosis: Do not claim to know a person's feelings, personality, health, or ability.
- Design, not the person: Measure how a design supports a task. Do not score the user.
- Choice and opt-out: Users can change, stop, correct, or refuse an adaptive feature.
- Minimum data: Use synthetic or anonymous task data and collect only what the activity needs.
- No hidden recording: Do not secretly collect voice, face, account, message, or sensor data.
- Accessible alternatives: Offer more than one way to see, hear, control, and complete a task.
Optional research extension
Sessions 13-16: deeper research.
Improve the pilot, test robustness, write a paper-style report, and check public reproducibility.
근거로 AI 프로그램 한 번 더 개선하기
Students complete Evidence-Led System Revision, make one clear design choice, test it, and record what changed.
Deeper Analysis and Robustness
Students complete Deeper Analysis and Robustness, make one clear design choice, test it, and record what changed.
다른 사람도 따라 할 수 있는 연구 보고서 쓰기
Students complete Research Paper Studio, make one clear design choice, test it, and record what changed.
공개 시연으로 결과와 한계 다시 확인하기
Students complete Public Demo and Reproducibility Review, make one clear design choice, test it, and record what changed.
Family FAQ
Before starting
this level.
Clear answers about tools, outputs, and privacy for grades 9-12.
Does my student need coding or design experience?
No required tool experience. We choose a suitable path from Figma, surveys, and spreadsheets, Python and Jupyter, Synthetic or public data, Optional non-sensitive sensors and keep the same design question and evidence standard.
What does a student make in each class?
Each class leaves visible work, such as observation notes, a requirement, a prototype, test data, a chart, or a before-and-after improvement.
Does a student need to share private feelings?
No. Personal sharing is optional. Every core activity has a neutral Nova School scenario and an anonymous recording option.
How can this course support a future college application?
The course was designed from open research published by Carnegie Mellon and Stanford. Students first explain how university researchers study whether technology is useful, easy, fair, and safe. They then distinguish their own question, design choice, A/B result, failure, revision, and limit for an authentic college application or interview explanation. A course title alone is not admissions evidence, and no course guarantees admission.
Can students take this course online or in person, one-to-one or in a small group?
Yes. The same roadmap is available online or in person at CIT in Apgujeong, Seoul, in one-to-one and small-group formats.
Course consultation
Turn a student's interest
into an experience project.
We can match the grade, coding experience, and topic to paper, Figma, Scratch, web, Python, or robotics.