What is the standard for Learning Income?
Machine-translated from Korean by Claude Haiku 5.5.Show original
학습소득의 기준은 무엇인가?
Doni's answer
The 'learning' that will serve as the basis for Learning Income is: through what experience did that person come to acquire knowledge that expanded beyond the knowledge and beliefs they originally held? A practical standard for measuring this is how much the expected correct-answer rate has increased on questions that the person, at the level of knowledge they were presumed to have, would likely not have been able to answer. Currently, measuring this accurately is nearly impossible.
The technique I am preparing for this is a methodology that extends Knowledge Tracing so that it can handle diverse contexts and types of interaction beyond standardized problem-solving records.
Meanwhile, to judge whether one has 'gained useful knowledge,' which is what this Learning Income would measure, we need to define useful knowledge, and that definition should emerge from social consensus. For now, I think the discussions on revising the national curriculum come closest to this process.
Machine-translated from Korean by Claude Haiku 5.5.Show original
학습 소득의 기준이 될 '학습'은 어떤 경험을 통해 그 사람이 원래 갖고 있던 지식과 신념보다 확장된 지식을 얻게 되었는가? 이다. 이를 실용적으로 측정할 수 있는 기준은, 그 사람이 갖고 있었을 것으로 여겨지는 지식 수준에서 대답할 수 없었을 것이라 여겨지는 질문의 예상 정답률이 얼마나 증가했는가이다. 이는 현재 정확하게 측정하기 거의 불가능하다.
내가 이를 위해 준비하려는 기술은 Knowledge Tracing이라는 방법론이 정형화된 문제 풀이 기록을 넘어 다양한 맥락과 상호작용의 종류에 대응할 수 있도록 확장하는 방법론이다.
한편, 이 학습소득에서 측정할 '유용한 지식을 얻었는가'를 판단하기 위해 유용한 지식을 정의해야 할텐데, 이는 사회의 합의에서 도출되어야 할 것이다. 현재로서는 교육과정 개정 논의가 이 과정에 가깝다고 생각한다.
Ask a follow-up to this answer
2 follow-up questions
- ↓US universities are turning away from AI-detection software because detectors often produce false positives that erode trust between instructors and students. When AI can write the assignment, how should we measure what a student actually learned? Does this affect your idea of measuring learning through knowledge tracing?
- ↓The Next Web, covering an OECD survey, reports that 67% of US firms use software that sanctions poor performance, against 4% in four European countries and 1% in Japan, and traces the gap to regulation. As someone who researches how to measure what people know, where would you draw the line between using measurement to punish people and using it to help them?