Skip to main content
CIT.EXPERIENCE LAB
EN한국어

COLLEGE APPLICATION RESEARCH EVIDENCE · 고등

Human-Friendly Adaptive Study Assistant인간친화적 적응형 학습 도우미

Clear reasons, visible uncertainty, low mental load, and final user control

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.

Course roadmap
Research questionStudent A/B testEvidence and limitsOwn contribution
01 / User and context

A learner who wants to control workload, support, and challenge

Choosing support, reading a recommendation, doing the task, correcting the result, stopping adaptation, and deleting data

Emotional experienceThinking experienceAction experienceSocial experienceTime and context
02 / Problem

The system does not show why it made a recommendation, and the user cannot easily correct it.

Target experienceClear reasons, visible uncertainty, low mental load, and final user control

03 / Testable requirements

Turn a need into a clear system promise

  • 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.
04 / Alternatives

One variable, two versions

  • A: a rule-based assistant controlled by the user
  • B: a limited AI assistant that can abstain and ask a question
05 / Practical test

Check it with the same real task

Use synthetic inputs and a consent-based usability pilot. Compare understanding, appropriate reliance, correction, and recovery.

06 / Evidence

Use numbers and user comments

  • Task and explanation understanding
  • Uncertainty reading and appropriate reliance
  • Failure, correction, and opt-out success
AB

The chart values are synthetic examples for this course page, not student results.

07 / Redesign

Use the evidence to improve once

Fix the highest-risk failure first, then use a baseline and regression tests to find new errors.

08 / Ethics and accessibility

A good score is not the whole experience

No diagnosis, surveillance, or manipulation. The system card records intended use, data, errors, and language or context limits.

Human control

The user keeps final control.

clarifycorrectoverrideopt-outdelete

Separate low confidence from missing information, and abstain when needed.

09 / Student output

Research plan, system map, working prototype, analysis, system card, and research poster

FigmaPython/JupyterSpreadsheetStatic web

Limit: A small class pilot is early evidence. Record how people and contexts differ, then name the next question to test.

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.

Chat on KakaoCall 02-540-2922
Ask about the course