AI in Education — research, products, policies, and pedagogical debates around the use of artificial intelligence (including LLMs, adaptive systems, tutoring agents, and analytics) in learning environments from K-12 to higher education and lifelong learning.
- File names: lowercase, hyphens, no spaces (e.g.,
adaptive-learning-systems.md) - Every wiki page starts with YAML frontmatter (see below)
- Use
wikilinksto link between pages ([[page-slug]]or[[page-slug|display text]]) - When updating a page, always bump the
updateddate+time (see Frontmatter above) - Every new page must be added to
index.mdunder the correct section - Every action must be appended to
log.md(local-only, gitignored likeraw/— not committed; persists on the local repo for scan-complete anchors) - Two page types:
articles/<slug>.mdfor individual papers,concepts/<slug>.mdfor broad topics that synthesize multiple papers. An article belongs on a concept page's Connected Articles list; a concept page explains the concept itself, not any single paper. - Provenance markers: On pages that synthesize 3+ sources, append
^[raw/papers/source-file.md]at the end of paragraphs whose claims come from a specific source. This lets a reader trace each claim back without re-reading the whole raw file. Optional on single-source pages where thesources:frontmatter is enough.
Only these fields are validated by the live schema (src/content.config.ts). Extra fields in
frontmatter are ignored at build time, so keep to this list.
---
title: Page Title
created: "YYYY-MM-DDTHH:MM:SS±HH:MM"
updated: "YYYY-MM-DDTHH:MM:SS±HH:MM"
type: article | concept
tags: [from taxonomy below]
sources: [raw/papers/source-name.md] # articles only
confidence: high | medium | low # how well-supported the claims are
---created / updated MUST be full quoted date+time timestamps (e.g. "2026-08-16T20:47:13-04:00"), never bare dates. The right sidebar and RSS sort by these via string comparison, and unquoted ISO timestamps shift to UTC (next day). created = wiki ingestion date (not paper pub date). Bump updated on any significant body edit and rebuild so the sidebar refreshes. Pages display date-only via .split('T')[0].
confidence is optional (defaults to medium) but recommended for opinion-heavy or fast-moving
topics. A confidence: low page signals weak or partial support so weak claims don't silently
harden into accepted wiki fact. There is no contested or contradictions field — the schema
does not validate them, so do not add them to frontmatter. Record unresolved disagreements in the
page body instead (see Update Policy below).
> **Synthesis:** 2-4 sentence summary of what the paper does and finds
## Key Findings
- 3-5 distinct bullet points with the most important results
## Connected Concepts
- [[concept-slug]] (3-6 genuinely related concepts)
## Connected Articles
- [[article-slug]] (2-4 genuinely related articles)
## Citation
Author, A. (2026). [*Full Title*](https://doi.org/...). Journal.
Exactly one ## Citation per article, as a single APA-style line (hyperlinked full title, first 6
authors + ", et al." for longer lists, source suffix per the wiki's citation rules).
> **Synthesis:** definition + overview of the concept
## Key research themes
Theme descriptions with [[wikilinks]] to related articles
## Connected Concepts
- [[concept-slug]]
## Connected Articles
- [[article-slug]]
(No citation section — concepts synthesize multiple sources.)
Raw sources ALSO get a small frontmatter block so re-ingests can detect drift:
---
source_url: https://example.com/article # original URL, if applicable
ingested: YYYY-MM-DD # or ingested_date
sha256: <hex digest of the raw content below the frontmatter>
---raw/ is gitignored and never committed. Raw files may also carry an authors: list (used as the
source of truth when a page's citation authors need verification). The sha256: lets a future
re-ingest of the same URL skip processing when content is unchanged, and flag drift when it has
changed. Compute over the body only (everything after the closing ---), not the frontmatter
itself.
Canonical tag list (reconciled 2026-08-03: 23 near-duplicate tags consolidated; taxonomy now indexes all 122 tags in use).
llm,generative-ai,adaptive-learning,intelligent-tutoring,edtech-platform,learning-analytics,automated-grading,ai-detection,agentic-ai,multi-agentreinforcement-learning,multimodal,prompt-engineering,knowledge-tracing,student-modeling,nlp-education,programming-its,architecture,interpretability,ai-generated-content,content-quality,dot-framework,validate-then-generate,verification
personalized-learning,formative-assessment,feedback-loop,scaffolding,active-learning,blended-learning,mastery-learning,educational-theorypedagogy,instructional-design,curriculum-design,assessment,authentic-assessment,portfolio-assessment,socratic-method,project-based-learning,problem-based-learning,collaborative-learning,collaborative-ai-tutoring,design-thinking,training-methodology,simulation,oral-defense,pair-programming
metacognition,self-regulated-learning,self-directed-learning,cognitive-offloading,critical-thinking,creative-thinking,computational-thinking,skill-decay,confidence,student-ai-interaction
k-12,higher-ed,lifelong-learning,teacher-role,student-experience,administrator,policy-makerfaculty-development,public-sector,practitioner-beliefs,educator-guide
stem-education,language-learning,writing-education,special-education,professional-trainingai-education,cs-education,math-education,physics-education,health-education,mooc,online-education,online-learning,distance-learning,blended-learning,multilingual-learning,neurodiversity,inclusive-learning,informal-learning,independent-learning,software-engineering,business-education,economics-education,management-education
rct,efficacy-study,benchmark,learning-gains,engagement-metrics,dropout-reduction,scoping-review,systematic-reviewai-ed-evaluation,assessment-validity,literature-review,meta-analysis,survey,qualitative-research,research-methods,evidence,ground-truth-reliability-aied
academic-integrity,bias-mitigation,privacy,equity,regulation,plagiarism-detection,hallucination-risk,over-reliance,remote-proctoring,automated-proctoringethics,pedagogical-safety,misinformation,global-south,culturally-sustaining-pedagogy,institutional-change
affective-computing,ai-literacy,human-in-the-loop,boundary-object,posthumanist,trust-calibration
comparison,timeline,controversy,prediction,market-analysis,open-source,stub
Rule: every tag on a page must appear in this taxonomy. If a new tag is needed, add it here first, then use it. This prevents tag sprawl.
- Create a page when an entity/concept appears in 2+ sources OR is central to one source
- Add to existing page when a source mentions something already covered
- DON'T create a page for passing mentions, minor details, or things outside the domain
- Split a page when it exceeds ~200 lines — break into sub-topics with cross-links
- Archive a page when its content is fully superseded — remove it from
index.mdand the live site
One page per concept or topic. Include:
- Definition / explanation
- Current state of knowledge
- Open questions or debates
- Related concepts (wikilinks)
When new information conflicts with existing content:
- Check the dates — newer sources generally supersede older ones
- If genuinely contradictory, note both positions with dates and sources in the page body
- If the contradiction is significant, lower the page's
confidencetomediumorlow - Flag for user review in the lint report
- Reverse chronological index of all ingested articles/papers.
- Automatically regenerated: sorted by
createddate in article frontmatter, newest first. - Shows: confidence icon (●/◐/○), wikilink to article page, source reference, full title, and tags.
- Excludes low-confidence stubs with no sources.
- Regenerate after every ingestion batch to keep current.