AI models trained on MOOC completion data risk bias toward high-performing learners, and machine-learned assessment reflects interaction rather than learning¶
Claim · All claims
Evidence · 1 study (2 entries) · theoreticalr2·q1
Currency · newest study 2017 · not yet checked against its sources · not reviewed by a person
Subclaims¶
Reading the evidence codes
q is the design tier, i the impact, n the sample, and the kind with its r
the rigour judged within that kind: what each code means.
This claim: no standardized effect is recorded in observations/ for it yet, so no benchmark reading is possible.
q1 i? Learning behaviour data are mostly from MOOC completers, so AI models trained on such data may be biased towards high performing learners. → Yu 2017
q1 i? Machine learning-based learner assessment using current MOOC interaction data mainly reflects how students interact with, rather than learn from, the courses. → Yu 2017 (2)
Evidence¶
Yu 2017¶
Yu, H., Miao, C., Leung, C. and White, T. J. (2017). Towards AI-powered personalization in MOOC learning. npj Science of Learning. https://doi.org/10.1038/s41539-017-0016-3
q1 · i? · theoretical · r2
The authors' analytical argument in the personalization section: because available behaviour data come mostly from completers, trained AI models may be biased towards high performing learners; they state the extent of this bias "requires more studies to be fully understood."
"as the available learning behaviour data are mostly from people who managed to complete the MOOCs, it is possible that the AI models trained from such data may be biased towards high performing learners."
Yu 2017 (2)¶
Yu, H., Miao, C., Leung, C. and White, T. J. (2017). Towards AI-powered personalization in MOOC learning. npj Science of Learning. https://doi.org/10.1038/s41539-017-0016-3
q1 · i? · theoretical · r2
The authors argue from existing MOOC analytics literature that, using data derived from interaction mechanisms currently available in MOOC platforms, learner assessment "mainly re flect how students interact with, rather than learn from, the courses".
"machine learning- based learner assessment techniques mainly re flect how students interact with, rather than learn from, the courses."
Discussion¶
Related Claims¶
- Algorithmic bias and data privacy are central ethical risks of AI in adult education — a broader claim this one bears on
- Behavioural engagement indicators such as platform interaction and task completion do not by themselves establish cognitive engagement with mathematical ideas — related
- A student model that is easily ported to another ITS may still fail there, because student behaviour depends on domain, interface, and the system that generated training data, so ported models must be re-tested. — related
Page metadata
| Field | Value |
|---|---|
type |
claim |
id |
mooc-ai-bias-toward-completers-and-interaction-not-learning |
title |
AI models trained on MOOC completion data risk bias toward high-performing learners, and machine-learned assessment reflects interaction rather than learning |
description |
AI models trained on MOOC completion data risk bias toward high-performing learners, and machine-learned assessment reflects interaction rather than learning |
status |
review |
evidence_strength |
weak |
newest_evidence |
2017 |
generated |
by process:wiki-ingest on 2026-10-27 |
trust |
unverified — no one has confirmed this page's content |
sources |
2 — yu-2017, yu-2017-2 |
Frontmatter is the page's machine-readable record — see the Schema & Guide.