From AI Literacy to AI Fluency: A Developmental Framework for Higher Education

May combine two claims.

Reviews of measurement instruments find good internal consistency but persistent gaps in content validity and performance-based assessment.

“Reviews of measurement instruments find good internal consistency but persistent gaps in content validity and performance-based assessment”

• cites Lintner, 2024 sourceSupported

Classified as probably not a substantial claim.

AI literacy is a developmental capacity rather than mere tool adoption.

“emerging work argues that AI literacy is a developmental capacity rather than mere tool adoption”

• Also stated in: § Developmental Sequencing and Longitudinal Study · plain

  “Liu and Levy (2026), in an emerging preprint, argue that AI literacy is a developmental capacity rather than mere tool adoption and report an observational case engaging more than 330 participants.”

• cites Liu & Levy, 2026 sourceSupported
• cites Liu and Levy, 2026 sourceSupported

Probably not a substantial claim: may not pass S2 (checkable).

Assessment policy should attend to enforcement feasibility and authentic learning rather than simple prohibition.

“Corbin et al. (2025) similarly argue that assessment policy should attend to enforcement feasibility and authentic learning rather than simple prohibition”

• Also stated in: § Assessment rubrics and policy alignment · plain

  “an approach consistent with Corbin et al. (2025), who argue that assessment policy should address authentic learning and enforcement feasibility rather than simple prohibition.”

• cites Corbin et al., 2025 sourceSupported

Classified as probably not a substantial claim.

AI and data competencies can be defined across four proficiency levels and seven knowledge dimensions.

“Kennedy and Gupta (2025) define AI and data competencies across four proficiency levels and seven knowledge dimensions, a structure that maps readily onto a level-by-domain matrix.”

• cites Kennedy and Gupta, 2025 sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

Chee et al. propose a lifelong AI literacy pathway with implications for curriculum design.

“Chee et al. (2024) propose a lifelong AI literacy pathway with implications for curriculum design”

• Also stated in: § Curriculum mapping and core learning outcomes · plain

  “Chee et al. (2024) propose a lifelong AI literacy development pathway with implications for curriculum design”

• cites Chee et al., 2024 sourceSupported

Classified as probably not a substantial claim. May combine two claims.

El-Banna et al. call for tailored faculty development alongside standardized AI education frameworks.

“El-Banna et al. (2025) call for tailored faculty development alongside standardized AI education frameworks”

• cites El-Banna et al., 2025 sourceSupported

May combine two claims.

Existing AI literacy scales show good structural validity yet gaps in content validity, cross-cultural validity, and performance-based assessment.

“Lintner (2024) found that existing AI literacy scales show good structural validity yet gaps in content validity, cross-cultural validity, and performance-based assessment.”

• cites Lintner, 2024 sourceSupported

Classified as probably not a substantial claim.

AI literacy (understanding and evaluating) is distinct from AI fluency (creating and innovating).

“Rogers and Carbonaro (2025) distinguish AI literacy (understanding and evaluating) from AI fluency (creating and innovating), reinforcing that outcome statements at the upper levels must require strategic creation and disciplinary application, not recognition alone.”

• cites Rogers and Carbonaro, 2025 sourceSupported

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

Such responses are necessary, but they mistake the nature of the challenge.

“Such responses are necessary, but they mistake the nature of the challenge.”

May combine two claims.

Most frameworks describe what AI literacy contains rather than how it develops, offering dimensional inventories without a validated progression from awareness to expert, responsible practice.

“Most frameworks describe what AI literacy contains rather than how it develops, offering dimensional inventories without a validated progression from awareness to expert, responsible practice.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

AI fluency is a socio-technical capacity rather than an individual skill alone, depending on curriculum design, assessment redesign, faculty development, equitable access, policy, data governance, and human-centered principles.

“The central claim is that AI fluency is not only an individual skill but a socio-technical capacity dependent on curriculum design, assessment redesign, faculty development, equitable access, institutional policy, data governance, academic-integrity expectations, privacy safeguards, and human-centered AI principles.”

Classified as probably not a substantial claim.

In an AAC&U and Elon survey, 95% of faculty said generative AI will increase students' overreliance on these tools.

“95% of faculty said generative AI will increase students' overreliance on these tools”

• Also stated in: § Faculty AI Competency · plain

  “95 percent of faculty expect generative AI to increase students' overreliance on these tools”

• cites Watson et al., 2026Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

Higher education has begun adapting general literacy constructs to its own context, yet they tend to stop at functional understanding and applied use.

“Read together, these works confirm that higher education has begun adapting general literacy constructs to its own context, yet they tend to stop at functional understanding and applied use.”

May combine two claims.

The frameworks proliferate without converging on a shared progression, leaving little common architecture for moving learners from awareness toward strategic, disciplinary, and leadership-level use.

“Second, the frameworks proliferate without converging on a shared progression, leaving little common architecture for moving learners from awareness toward strategic, disciplinary, and leadership-level use.”

Classified as probably not a substantial claim.

Across accounts, literacy describes a baseline of comprehension and competent, evaluative use.

“Across these accounts, literacy describes a baseline of comprehension and competent, evaluative use.”

May combine two claims.

In the AAC&U and Elon University survey, 90 percent of faculty said generative AI use will diminish students' critical thinking skills and 69 percent reported having addressed AI literacy issues in their institutions.

“in the AAC&U and Elon University survey, 90 percent of faculty said generative AI use will diminish students' critical thinking skills and 69 percent reported having addressed AI literacy issues in their institutions (Watson et al., 2026).”

• Also stated in: § Assessment rubrics and policy alignment · plain

  “with faculty in the AAC&U and Elon survey expressing concern that the tools will increase overreliance and diminish critical thinking (Watson et al., 2026)”

• cites Watson et al., 2026Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

Literacy supplies the conceptual and evaluative grounding without which strategic use degenerates into uncritical reliance, and fluency converts that grounding into adaptive practice that transfers across tasks and disciplines.

“Literacy supplies the conceptual and evaluative grounding without which strategic use degenerates into uncritical reliance, and fluency converts that grounding into adaptive practice that transfers across tasks and disciplines.”

Probably not a substantial claim: may not pass S2 (checkable).

Without a disciplined distinction, institutions risk treating exposure to tools as though it were mastery of practice.

“Without a disciplined distinction, institutions risk treating exposure to tools as though it were mastery of practice.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

Exposure alone confers no understanding, which is why a developmental pathway must begin below awareness.

“Exposure alone confers no understanding, which is precisely why a developmental pathway must begin below awareness.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

Across AI competency frameworks, there is persistent divergence in the number, naming, and progression of levels, and most progression schemes remain shallow, with three levels being common.

“Second, there is persistent divergence in the number, naming, and progression of levels, and most progression schemes remain shallow, with three levels being common.”

May combine two claims.

In the United Kingdom, 95 percent of students use AI in at least one way and 94 percent use generative AI to help with assessed work.

“The HEPI and Kortext survey of United Kingdom students reports that 95 percent of students use AI in at least one way and 94 percent use generative AI to help with assessed work (Stephenson et al., 2026).”

• Also stated in: § Assessment rubrics and policy alignment · plain

  “With 95% of students reporting AI use in at least one way and 94% using generative AI for assessed work (Stephenson et al., 2026)”

• cites Stephenson et al., 2026Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

AI literacy is treated as the organizing competency, but the construct remains unsettled.

“AI literacy is treated as the organizing competency, but the construct remains unsettled.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

Ethical and integrity concerns are central and recurring in the literature, yet it largely treats ethics as a category of risk to be recognized rather than as a graduated competency that learners develop and exercise with growing sophistication.

“Together this strand demonstrates that ethical and integrity concerns are central and recurring, yet it largely treats ethics as a category of risk to be recognized rather than as a graduated competency that learners develop and exercise with growing sophistication.”

Classified as probably not a substantial claim. May combine two claims.

Literacy equips a learner to recognize what AI is, how it works, where it fails, and what risks it carries, but it does not, by itself, specify how a learner deploys AI strategically within a discipline or workflow.

“It equips a learner to recognize what AI is, how it works, where it fails, and what risks it carries, but it does not, by itself, specify how a learner deploys AI strategically within a discipline or workflow.”

May combine two claims.

Much of the framework literature centers students, yet faculty hold the pedagogical and assessment levers that determine whether literacy or fluency is cultivated, and faculty themselves anticipate substantial harms.

“This strand reveals a critical asymmetry: much of the framework literature centers students, yet faculty hold the pedagogical and assessment levers that determine whether literacy or fluency is cultivated, and faculty themselves anticipate substantial harms.”

May combine two claims.

The literature treats individuals, faculty, and institutions in relative isolation, even as workforce and governance evidence shows these levels are interdependent.

“Third, the literature treats individuals, faculty, and institutions in relative isolation, even as workforce and governance evidence shows these levels are interdependent.”

Probably not a substantial claim: may not pass S2 (checkable).

Fluency is the strategic, accountable, and discipline-aware use of AI in which human judgment, authorship, and standards are preserved.

“Fluency, in this account, is the strategic, accountable, and discipline-aware use of AI in which human judgment, authorship, and standards are preserved.”

Classified as probably not a substantial claim.

AI fluency is named but not yet theorized as a developmental framework for higher education.

“AI fluency, in short, is named but not yet theorized as a developmental framework for higher education.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

AI literacy describes awareness, functional use, and recognition of broad risks such as bias and hallucination; it does not, by itself, describe the strategic, critical, ethical, and disciplinary capacities required to use generative AI as a genuine partner in scholarly and professional work.

“AI literacy describes awareness, functional use, and recognition of broad risks such as bias and hallucination; it does not, by itself, describe the strategic, critical, ethical, and disciplinary capacities required to use generative AI as a genuine partner in scholarly and professional work.”

Probably not a substantial claim: may not pass S3 (load-bearing).

A larger proportion of institutions now view AI as a strategic priority, at 57 percent compared with 49 percent the prior year.

“The EDUCAUSE landscape similarly reports that a larger proportion of institutions now view AI as a strategic priority, at 57 percent compared with 49 percent the prior year (Robert et al., 2025).”

• cites Robert et al., 2025Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S2 (checkable). May combine two claims.

Tiered AI literacy courses paired with continuous curriculum updating are needed to keep pace with tool change.

“Ma et al. (2025) argue for tiered AI literacy courses paired with continuous curriculum updating to keep pace with tool change.”

• cites Ma et al., 2025Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim. May combine two claims.

The NIST AI Risk Management Framework organizes trustworthy AI around the GOVERN, MAP, MEASURE, and MANAGE functions and specifies characteristics such as accountability, transparency, and privacy enhancement.

“The NIST AI Risk Management Framework organizes trustworthy AI around the GOVERN, MAP, MEASURE, and MANAGE functions and specifies characteristics such as accountability, transparency, and privacy enhancement (National Institute of Standards and Technology (NIST), 2023).”

• cites National Institute of Standards and Technology (NIST), 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

AI and big data are the fastest-growing skills, and 39 percent of workers' existing skill sets will be transformed or become outdated over the 2025 to 2030 period.

“The World Economic Forum identifies AI and big data as the fastest-growing skills and projects that 39 percent of workers' existing skill sets will be transformed or become outdated over the 2025 to 2030 period (World Economic Forum, 2025).”

• cites World Economic Forum, 2025Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

A growing share of institutions treat AI as a strategic priority, yet capacity is uneven.

“Robert et al. (2025) report that a growing share of institutions treat AI as a strategic priority, yet capacity is uneven.”

• cites Robert et al., 2025Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim.

Chan offers an AI Ecological Education Policy Framework spanning pedagogical, governance, and operational dimensions.

“Chan (2023) offers an AI Ecological Education Policy Framework spanning pedagogical, governance, and operational dimensions, providing a structure within which outcome statements, rubrics, and data governance can be reconciled.”

• cites Chan, 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing).

Song et al. introduce the FALCON-AI Scale for faculty across teaching, research, and service.

“in an emerging preprint, Song et al. (2026) introduce the FALCON-AI Scale for faculty across teaching, research, and service.”

• cites Song et al., 2026 sourceNot checked — no full text of the source

May combine two claims.

Skills are central to realizing AI's value, and skills gaps are a leading barrier to adoption.

“The OECD brief synthesizes evidence that skills are central to realizing AI's value and identifies skills gaps as a leading barrier to adoption (OECD, 2026).”

• cites OECD, 2026Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Artificial intelligence can reshape education systems through curriculum integration, personalized learning, and administrative transformation.

“The World Economic Forum (2024) further argues that artificial intelligence can reshape education systems through curriculum integration, personalized learning, and administrative transformation”

• cites World Economic Forum, 2024Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing).

The ethical issues most prominent in higher education concern academic integrity.

“In an emerging preprint, Li et al. (2023) report that the ethical issues most prominent in higher education concern academic integrity.”

• cites Li et al., 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S4 (based).

AI fluency advances only when curriculum, assessment, faculty development, and governance move together.

“Because AI fluency is a socio-technical capacity, it advances only when curriculum, assessment, faculty development, and governance move together.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

UNESCO guidance frames the goal as ensuring a human-centred vision of generative AI and warns that the absence of regulation leaves user data privacy unprotected and institutions unprepared.

“UNESCO guidance frames the goal as ensuring a human-centred vision of generative AI and warns that the absence of regulation leaves user data privacy unprotected and institutions unprepared (Miao et al., 2023).”

• cites Miao et al., 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

AI shifts educational technology toward automating decisions about instruction, which raises the stakes for institutional oversight.

“The United States Department of Education emphasizes that AI shifts educational technology toward automating decisions about instruction, which raises the stakes for institutional oversight (U.S.Department of Education, Office of Educational Technology, 2023).”

• cites U.S.Department of Education, Office of Educational Technology, 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S2 (checkable).

AI fluency is a socio-technical capacity.

“Because AI fluency is a socio-technical capacity, it advances only when curriculum, assessment, faculty development, and governance move together.”

• Also stated in: § Discussion · plain

  “It is, moreover, a socio-technical capacity that depends on curriculum and assessment redesign, faculty development, equitable access, institutional policy, data governance, and human-centered principles (Miao et al., 2023;Chan, 2023;National Institute of Standards and Technology, 2023).”

Probably not a substantial claim: may not pass S3 (load-bearing).

Song et al. advance a faculty-specific instrument spanning teaching, research, and service.

“In an emerging preprint, Song et al. (2026) advance a faculty-specific instrument spanning teaching, research, and service, underscoring that role-embedded competencies differ from those of students or the general public.”

• cites Song et al., 2026 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Student competencies progress through Understand, Apply, and Create levels.

“Miao et al. (2024) describe student competencies progressing through Understand, Apply, and Create levels, which align with the framework's literacy band.”

• cites Miao et al., 2024Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

AI literacy, while necessary, is no longer sufficient for higher education in the generative AI era.

“This article has argued that AI literacy, while necessary, is no longer sufficient for higher education in the generative AI era.”

Probably not a substantial claim: may not pass S2 (checkable). May combine two claims.

Framework adoption must be anchored in human-centered principles and equitable access, or it will widen the gaps it intends to close.

“Framework adoption must be anchored in humancentered principles and equitable access, or it will widen the gaps it intends to close.”

Classified as probably not a substantial claim.

The developmental framework reframes academic integrity from a policing problem into a competency outcome.

“The developmental framework reframes integrity from a policing problem into a competency outcome”

Probably not a substantial claim: may not pass S4 (based).

Faculty development is the binding constraint on framework adoption.

“Faculty development is the binding constraint on framework adoption.”

Probably not a substantial claim: may not pass S2 (checkable). May combine two claims.

Foundational courses can carry Conceptual AI Understanding and Operational AI Skill (Levels 1 and 2), while capstone and disciplinary courses carry Epistemic Verification, Disciplinary and Professional Transfer, and the higher fluency bands (Levels 4 through 6).

“Foundational courses can carry Conceptual AI Understanding and Operational AI Skill (Levels 1 and 2), while capstone and disciplinary courses carry Epistemic Verification, Disciplinary and Professional Transfer, and the higher fluency bands (Levels 4 through 6).”

Classified as probably not a substantial claim.

AI fluency is expressed through disciplinary standards rather than in the abstract.

“A fourth direction is discipline-specific case research, because fluency is expressed through disciplinary standards rather than in the abstract.”

Probably not a substantial claim: may not pass S2 (checkable). May combine two claims.

AI fluency depends on curriculum and assessment redesign, faculty development, equitable access, institutional policy, data governance, and human-centered principles.

“It is, moreover, a socio-technical capacity that depends on curriculum and assessment redesign, faculty development, equitable access, institutional policy, data governance, and human-centered principles (Miao et al., 2023;Chan, 2023;National Institute of Standards and Technology, 2023).”

• cites Miao et al., 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Chan, 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites National Institute of Standards and Technology, 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

No general education or assessment reform succeeds without prepared educators, and faculty readiness remains a binding constraint.

“No general education or assessment reform succeeds without prepared educators, and faculty readiness remains a binding constraint.”

Probably not a substantial claim: may not pass S2 (checkable).

Effective governance binds policy, data governance, privacy safeguards, and academic-integrity expectations to the competency model so that Level 6 Responsible AI Leadership becomes an institutional achievement rather than merely an individual aspiration.

“Effective governance binds policy, data governance, privacy safeguards, and academic-integrity expectations to the competency model so that Level 6 Responsible AI Leadership becomes an institutional achievement and not merely an individual aspiration.”

May combine two claims.

The fluency band connects directly to graduate outcomes because employers increasingly treat AI capability as foundational.

“The fluency band connects directly to graduate outcomes, because employers increasingly treat AI capability as foundational.”

Probably not a substantial claim: may not pass S2 (checkable).

AI fluency is the capacity to think, learn, create, evaluate, and act responsibly with AI while preserving human judgment, authorship, accountability, and disciplinary standards.

“It is the capacity to think, learn, create, evaluate, and act responsibly with AI while preserving human judgment, authorship, accountability, and disciplinary standards (Long & Magerko, 2020).”

• cites Long & Magerko, 2020 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S2 (checkable).

Preparing graduates for an AI-saturated work environment requires the higher tiers of the framework, where Operational AI Skill and Disciplinary and Professional Transfer mature into workflow design and judgment under real conditions.

“Preparing graduates for this environment requires the higher tiers of the framework, where Operational AI Skill and Disciplinary and Professional Transfer mature into workflow design and judgment under real conditions.”

Probably not a substantial claim: may not pass S4 (based).

Across its six competency domains, the framework treats conceptual understanding, operational skill, epistemic verification, ethical and legal judgment, metacognitive and human agency, and disciplinary transfer as interdependent capacities rather than isolated skills.

“Across its six competency domains, the framework treats conceptual understanding, operational skill, epistemic verification, ethical and legal judgment, metacognitive and human agency, and disciplinary transfer as interdependent capacities rather than isolated skills.”

• AssumptionRests on an assumption: The six competency domains (Conceptual AI Understanding, Operational AI Skill, Epistemic Verification, Ethical and Legal Judgment, Metacognitive and Human Agency, and Disciplinary and Professional Transfer) adequately capture the capacities that constitute AI fluency.

Classified as probably not a substantial claim.

The proposed seven-level framework, spanning Level 0 AI Exposure through Level 6 Responsible AI Leadership, positions literacy (Levels 0 to 3) as the necessary foundation on which fluency (Levels 4 to 6) is built.

“the proposed seven-level framework, spanning Level 0 AI Exposure through Level 6 Responsible AI Leadership, positions literacy (Levels 0 to 3) as the necessary foundation on which fluency (Levels 4 to 6) is built.”

• AssumptionThe authors state this as an assumption: The proposed seven-level developmental sequence (exposure through responsible AI leadership) is a valid ordering of learner development.