Cross-university platform portalShow a desktop notification when the AI TA finishes replying
Notify me when classmates post messages in the forum
Play an alert sound whenever there is a new notification
An integrated competency framework constructed by extracting common dimensions from six international and local AI literacy frameworks, proposing a 6 × 3 = 18 cell coordinate system as a design tool for AI literacy courses in higher education and for institutional self-assessment.
The cross-product of six dimensions × three levels forms 18 cells, serving as an AI literacy course design and self-assessment tool for higher education institutions
| Dimension | L1 Acquire Foundations | L2 Deepen Application | L3 Create Creation and Design |
|---|---|---|---|
| AI foundationsD1 | Know what AI is and what it can and cannot do | Understand how ML / LLM works | Can explain AI concepts to Students/peers |
| AI ethics and responsibilityD2 | Understand the basic issues in AI ethics | Apply ethical judgement in your own learning activities | Can promote discussion of ethical norms in the learning community and reflect on one's own practice |
| Data and privacy governanceD3 | Understand consent mechanisms and privacy risks | Has the ability to reflect on their own data trail | Can plan learning activities or peer-collaboration schemes that comply with privacy principles |
| Critical useD4 | Know that AI can make mistakes | Can verify/challenge AI outputs | Can design a systematic information verification process and demonstrate it to peers |
| Integrated application of AI in learning tasksD5 | Can use AI to complete simple learning tasks | Can choose the appropriate level of AI intervention according to the nature of the task | Can design a personalised AI-assisted learning workflow |
| Learner AutonomyD6 | Know that AI can be independently shaped or refused by learners | Can override AI recommendations and adjust AI behaviour to serve one's own learning | Use AI to create things that did not originally exist, design new AI applications or learning pathways |
The cell marked in yellow (D6 × L3 Create) is a conceptual blind spot identified by this framework — this position is under-operationalised across all six frameworks, and is a research and design issue that deserves priority attention from academia, policy circles and platforms in the generative AI era.
Advanced use: In addition to serving as a static self-assessment checklist, each Course can present its own literacy distribution profile across these 18 cells, makingcross-Course comparison, department-level aggregation, and inter-institution benchmarking possible. The full measurement methodology and cross-institution applications will be published in a companion paper.
Common capability/responsibility dimensions cross-extracted from six frameworks, with each dimension appearing in at least three source frameworks
Understand the nature of AI, how it works, and what it can and cannot do: from identifying AI systems and distinguishing general-purpose from specialised AI, to mastering the basic principles of machine learning and language models.
Understand and be able to apply AI ethics judgements: bias, fairness, explainability, and accountability. A core topic that covers all six frameworks.
Consent mechanism, reflection on data trails, and privacy boundary design: from the Personal Data Protection Act / GDPR concepts to platform deployment implementation.
Practical ability to verify, challenge, and detect issues in AI outputs, alongside — rather than replacing — the normative ethical thinking of D2.
Learners integrate AI within existing learning tasks to improve efficiency and effectiveness (augment learning): using AI as a scaffold and accelerator to complete assigned or self-set learning goals.
Learners cango beyond the task or reject AI suggestions (transcend or resist), autonomously shaping, adapting and designing AI systems to create things that did not originally exist. This framework identifies this dimension as under-operationalised across all six frameworks in the Create layer, representing a gap in the research agenda for the generative AI era.
Boundary between D5 and D6: both are learner-facing; the difference lies in the dominant action. D5 is augment learning (using AI within an existing task to do the learning better); D6 is transcend or resist (going beyond the task, creating something new, or rejecting AI suggestions). If an activity involves both, classify it by the dominant action: task-driven as D5, autonomous-driven as D6.
Four-layer argument marked as Contribution Finding for D6 (learner autonomy)
D1 (concept), D2 (ethics), D3 (data), D4 (critique), and D5 (integration) all have complete operationalisation across the six frameworks; only D6 (learner autonomy) exists in fragmented form — appearing in AAAI's C17 programmability, UNESCO CFS's AI system design, and the MOE's 'student subjectivity' statement — but it has never been elevated into a complete dimension spanning Acquire/Deepen/Create.
More precisely, D6 × L3 (Create) — learners use AI to create something that did not originally exist, override AI suggestions, and design new AI applications — this cell has no corresponding operational description in OECD, AAAI, IEEE 7000, the EU AI Act, MOE or UNESCO CFS. This is the concrete identification of a "shared blind spot across frameworks".
The core capability of Generative AI is to create. Without a D6 × L3 framework, learners are locked into the position of an "AI user" (the D5 augment orientation) and can never reach the position of an "AI author / agent" (the D6 transcend orientation). In the era of widespread GenAI, this position has shifted from a "future issue" to something that must be addressed now.
Contribution Finding is the claim that a reviewer will challenge directly: 'What has your framework added to the existing six frameworks?' This claim can be verified (the reviewer can check the six frameworks), can be falsified (if a framework actually has D6×L3, the contribution collapses), and has temporal relevance — all three are in place, so it is a publishable contribution rather than a literature review.
Conclusion: 6D Framework is not a literature review that “lines up six frameworks in a table”; rather, it is about identifying shared blind spots through systematic coding, and giving that blind spot actionable dimension naming and hierarchical descriptions. This is the contribution claim of Paper A, and the scholarly basis for D6 being marked with a Contribution Finding ribbon on the page and with D6 × L3 highlighted in yellow in the 18-Cell matrix.
Each dimension can be divided into three levels according to progress, designed in parallel with the UNESCO AI CFS, and derived from the upper three levels of the revised Bloom taxonomy by Anderson & Krathwohl 2001
Build basic knowledge and identification ability for this dimension; able to describe, define, and recognise relevant concepts and issues.
Can apply judgement and skills in that dimension within one's own learning context, and reflect on one's own practice.
Can design new schemes, demonstrate them to peers, and produce new tools or new learning schemes, involving a high degree of autonomy and creative expression.
Across intergovernmental principles, academic competitiveness frameworks, engineering standards, legislation, local policies, and international organisational competency frameworks
5 value-based principles + 5 national policy recommendations, leaning towards "system design principles", providing aligned language for platform-level ethical commitments.
5 key questions → 17 core competencies, the most complete framework for the individual capability dimension, and the de facto baseline in the AI literacy field.
The strongest anchor aligned at both the system and process levels, especially the draft IEEE 7004 student data governance framework, directly addressing research ethics for education platforms.
Most educational AI modules (outcomes assessment, exam monitoring) are classified as high risk; Article 4 AI literacy obligations are the bridge between individual capability and system deployment.
Transparent explanation, Student agency, multiple assessment, risk awareness, fairness and reasonableness, ethical responsibility — local policy alignment is the necessary proof of Uedu higher education user adoption.
CFS comprises 4 aspects × 3 levels = 12 blocks. This framework is positioned as learner-facing, corresponding only to CFS (Student version), not to CFT (Instructor version). The 3-level progression model is designed in parallel with CFS to facilitate cross-framework dialogue; CFS is under-specified in the D3 and D6 dimensions, and that is precisely the contribution of 6D in extending CFS.
AI Literacy 6D and Educational Omics are parallel rather than hierarchical — the former answers 'What kind of learner should learners become' (competency), while the latter answers 'What measurable dimensions are present in learners' (measurement).
Methodological decisions in the development of this framework, avoiding the trap of 'enumerative mapping'
The extraction of the six dimensions precedes access to Uedu platform data. Only after the dimensions are established is the platform used as instantiation evidence, to avoid the criticism that it was reverse-engineered from platform features.
Each common dimension must appear in at least 3 source frameworks, ensuring that the dimension is not a standalone extension of any one framework, but an inductive result of cross-framework consensus.
After open coding the six frameworks, perform axial coding; the codebook and inter-rater reliability will be disclosed in the Appendix of the formal paper to support reproducibility.
Within this framework, the Uedu platform is positioned as instantiation evidence, not the subject of analysis. For each dimension, 1–2 representative tools are selected; any uncovered cell is clearly labelled.
Two complementary frameworks make up a complete measurement-to-competency research programme (for deep readers, an anchor)
Dr. Chia-Kai Chang's research project is built around two complementary frameworks.Educational OmicsIt is an educational practice that operationalises Multimodal Learning Analytics (MMLA), putting the MMLA paradigm into practice in university general education settings and surfacing six observable dimensions: Cognomics, Linguomics, Physioneuromics, Sociomics, Environomics and Ethicomics.AI Literacy 6D Framework 是 learner-facing competency framework，整合 UNESCO AI CFS、OECD AI Principles、IEEE 7000 系列、EU AI Act、AAAI AI Literacy 與教育部生成式 AI 原則，抽取 D1 概念基礎、D2 倫理責任、D3 資料治理、D4 批判性使用、D5 學習任務整合（augment）、D6 創造與自主性（transcend）六向度與 Acquire／Deepen／Create 三層進展，回答「學習者應發展哪些 AI 素養」。兩框架在 Ethics 與 Agency 維度交會，並以 Uedu 平台（29 Deployments by higher education institutions,420,000+ 學習者—AI 互動，NCU 為 first instantiation）作為共同實證基礎，構成 measurement-to-competency 的完整 research programme。
About 220 words · Suitable for external presentations, research grant applications, and academic communication introductionsChia-Kai Chang's research programme rests on two complementary frameworks. Educational Omics (EO) is an educational practice that operationalizes Multimodal Learning Analytics (MMLA) in higher education general education contexts, with six observable dimensions—Cognomics, Linguomics, Physioneuromics, Sociomics, Environomics, and Ethicomics—emerging from sustained practice. The AI Literacy 6D Framework is a learner-facing competency framework synthesizing UNESCO AI CFS, OECD AI Principles, IEEE 7000 standards, the EU AI Act, AAAI AI Literacy, and Taiwan's MOE GenAI principles into six common dimensions (D1 conceptual foundations, D2 ethics, D3 data governance, D4 critical use, D5 AI integration in learning tasks—augment, D6 creation and learner agency—transcend) across three progression levels (Acquire, Deepen, Create), addressing what AI competencies learners should develop. The two frameworks intersect at the ethics and agency dimensions, and share Uedu Platform (deployed across 29 higher-education institutions, 420,000+ learner–AI interactions, with NCU as the first instantiation) as their common empirical foundation, together forming an integrated measurement-to-competency research programme.
~200 words · For international outreach, grant applications, academic correspondenceAcademic convention statement: This framework corresponds to Paper A and is currently at the pre-peer-review stage; it is a working draft and may be adjusted in response to review comments. When citing this page, please note “working draft, subject to revision”. The final version will be updated in step with publication of the paper.
The methodology decisions behind this framework’s corresponding paper (Paper A) are currently at the pre-peer-review stage. The formal version will be updated in step with the paper’s publication, and the institutional self-assessment tool will be made available for download.
Learn about Educational Omics View research ethics disclosure