Source-linked AI summary
AI Literacy in K-12 and Higher Education in the Wake of Generative AI: An Integrative Review
Xingjian Gu, Barbara J. Ericson
TL;DR
AI literacy has become prominent in education, but its definition remains vague across substantially different interventions and contexts. This paper conducts an integrative review of 124 studies and identifies a framework combining three literacy orientations with three perspectives on AI. The review finds major shifts in the field after generative AI and supports more specialized terminology while noting that the study does not evaluate specific research questions within AI literacy.
Problem
AI literacy lacks consensus, encompassing interventions that range from kindergarten social-robot activities to undergraduate ChatGPT use.
Method
The paper applies an integrative review method to synthesize 124 empirical and theoretical AI literacy studies.
Results
The review identifies three literacy orientations—functional, critical, and indirectly beneficial—and three AI perspectives—technical detail, tool, and sociocultural.
Takeaways & Limitations
The framework highlights the need for more specialized terms within AI literacy discourse and identifies research gaps in some AI literacy objectives.
Takeaways & Limitations
The integrative review generates a framework but does not evaluate specific research questions within AI literacy.
Abstract
from arXiv · showhide
Even though AI literacy has emerged as a prominent education topic in the wake of generative AI, its definition remains vague. There is little consensus among researchers and practitioners on how to discuss and design AI literacy interventions. The term has been used to describe both learning activities that train undergraduate students to use ChatGPT effectively and having kindergarten children interact with social robots. This paper applies an integrative review method to examine empirical and theoretical AI literacy studies published since 2020. In synthesizing the 124 reviewed studies, three ways to conceptualize literacy-functional, critical, and indirectly beneficial-and three perspectives on AI-technical detail, tool, and sociocultural-were identified, forming a framework that reflects the spectrum of how AI literacy is approached in practice. The framework highlights the need for more specialized terms within AI literacy discourse and indicates research gaps in certain AI literacy objectives.
1 Introduction
AI literacy has become an important but ambiguously defined educational objective, spanning technical understanding, tool use, critical evaluation, and sociocultural implications. The paper argues for a framework that organizes these diverse approaches and supports more precise communication.
- AI literacy definitions differ on whether they emphasize technical foundations, AI-tool use, or critical evaluation.
- AI literacy encompasses both functional skills for using AI technologies and understanding their broader sociocultural implications.
- Generative AI has further broadened the field by increasing access to capable language-model tools and prompting researchers to reconsider educational objectives.
- Because AI literacy is an umbrella term covering dissimilar interpretations, the paper proposes a framework to organize the field and enable clearer communication.
- The term covers substantially different interventions, from kindergarteners interacting with social robots to undergraduates practicing prompt engineering for large language models.
RQ1: How do researchers conceptualize and approach AI literacy in K-12 and higher education?
The first research question asks how researchers conceptualize and approach AI literacy in K-12 and higher education after generative AI.
- The review asks how researchers conceptualize and approach AI literacy in K-12 and higher education in the wake of generative AI.
generative AI?
The paper uses an integrative review of 124 studies to synthesize AI literacy research published since generative AI became prominent. It develops a unified framework and advocates more precise terminology for the field.
- The review analyzes 124 AI literacy studies published between 2020 and 2024.
- It examines the forms of AI addressed and the capabilities that studies seek to promote in students.
- The synthesis derives common conceptualizations and presents them in a unified framework.
- The paper updates prior reviews, delineates AI literacy’s boundaries, and adapts digital-literacy theories to build the framework.
- The framework situates existing competency lists within a broader theoretical landscape rather than replacing them.
- The paper advocates terms that describe researchers’ objectives and approaches more precisely than generic “AI literacy”.
2 Prior Work
Prior work treats AI literacy as a difficult-to-define construct shaped by changing AI technologies, multiple literacy traditions, and technical, user-oriented, and sociocultural perspectives.
- Long and Magerko’s influential framework defines AI literacy through competencies for evaluating, communicating and collaborating with, and using AI.
- Prior reviews proposed four AI-literacy aspects: knowing and understanding, using and applying, evaluating and creating, and ethical issues.
- Existing taxonomies are comprehensive but do not address educators’ divergent interpretations and resulting implementations of AI literacy.
- AI has shifted across technological paradigms, while older paradigms such as knowledge representation and reasoning are rarely mentioned in the reviewed studies.
- The review treats AI as a sociotechnical system with social, cultural, and political implications rather than only as a technical field.
- The framework draws on the Dagstuhl Triangle’s technical, user-oriented, and sociocultural perspectives for discussing digital systems.
- Literacy traditions distinguish functional knowledge, critical thinking and empowerment, and indirectly beneficial virtues as divergent motivations for literacy.
3 Integrative Review Methodology
The review addresses a vaguely defined and rapidly expanding AI literacy field containing diverse objectives, populations, and study types. It therefore uses an integrative review to synthesize theoretical and empirical literature and construct a broader conceptual framework.
- Rationale: AI literacy lacks a clear definition, while interventions pursue substantially different learning objectives.The literature includes both theoretical and intervention studies across fields, making the review landscape heterogeneous.
- Rationale: An integrative review was selected to combine theoretical and empirical literature and identify overarching patterns rather than adjudicate specific intervention outcomes.This method is described as suitable for emerging topics with vague boundaries and diverse research traditions.
- Search strategy: The database search produced 1,559 Scopus results and 44 ERIC results, which were reduced to 107 and 9 sources after screening and eligibility assessment.Eight additional studies were added through purposive searching, including one influential 2019 paper.
- Data extraction and analysis: Data extraction examined each study’s AI literacy definitions, theoretical frameworks, cited studies, motivations, goals, and research questions.The analysis began with bottom-up coding and iteratively refined themes before incorporating a more top-down theoretical structure.
4 Results
The review organizes AI literacy through three AI perspectives and three literacy perspectives. These categories can be combined freely, allowing the framework to represent diverse interventions without treating them as mutually exclusive.
- AI perspectives: The framework identifies AI as technical details, AI as tools, and AI’s sociocultural impact.Examples include teaching supervised learning algorithms, effective ChatGPT use, and bias in facial recognition.
- Framework structure: The three AI and three literacy conceptualizations can be freely combined and are not mutually exclusive.The framework therefore represents interventions through combinations of what AI is addressed and what learner capability or outcome is pursued.
- Literacy perspectives: The framework identifies functional, critical, and indirectly beneficial literacy perspectives.These correspond to preparation for AI-related work, informed citizenship around AI, and broader outcomes such as STEM interest.
- Framework structure: A curriculum combining machine-learning fundamentals with algorithmic-bias analysis exemplifies technical-detail and sociocultural AI perspectives alongside functional and critical literacy perspectives.The example links proficiency with critical awareness of AI technologies.
4.2 Overview of Reviewed Studies
The reviewed literature expanded rapidly and became increasingly focused on higher education, generative-AI tools, and indirect benefits, while combinations involving critical tool use remained comparatively underrepresented. The review thus reveals both dominant approaches and gaps in the AI literacy landscape.
- Scope and diversity: The review synthesized 124 studies, including empirical work on LLM-and-prompt literacy, middle-school curricula, and machine learning for college non-CS students.The diversity of these studies illustrates the complexity of AI literacy as a concept.
- Educational context: Studies in post-secondary contexts increased much more sharply than studies in K-12 contexts.The comparison concerns the educational-context trends shown across the review period.
- AI perspectives: Studies treating AI as tools increased rapidly from 2023 and exceeded technical-detail and sociocultural perspectives by July 2024.The review attributes this pattern to studies teaching the use of generative AI tools.
- Literacy perspectives: Indirect-benefit perspectives emerged from 2023, including STEM interest, computational thinking, and intrinsic motivation for studying computer science.Functional and critical literacy perspectives remained comparatively similar across the review period.
- Perspective combinations: Among 88 functional-literacy studies, 63 also used the AI technical-detail perspective.Common pairings also included sociocultural AI with critical literacy and AI tools with functional literacy.
- Research gaps: Relatively few studies combined an AI-tool perspective with critical literacy or an AI sociocultural perspective.This pattern indicates a research gap concerning responsible and critical use of AI.
4.3 Established AI Literacy Conceptualizations
Before generative AI, AI literacy studies used varied curricula and objectives across K-12 and post-secondary settings. The reviewed work spans functional and critical literacy, alongside technical-detail and sociocultural perspectives on AI.
- Six studies framed AI literacy functionally through machine-learning knowledge and skills, including implementing supervised-learning algorithms for computer-vision tasks.
- Purely technical machine-learning curricula appeared in K-12 settings, while comparable college courses were rarely labeled AI literacy.
- Ten studies treated AI as sociotechnical systems whose sociocultural influences should be understood by learners.
- Critical-literacy studies examined AI’s social applications, stakeholders, design intentions, accountability, and fairness.
- Twenty-seven percent of reviewed studies combined functional and critical literacy objectives, including machine-learning knowledge, AI-career interest, and ethical concerns.
- Post-secondary studies without generative AI were relatively rare and included three empirical intervention studies for undergraduate non-CS students.
4.4 Emerging AI Literacy Conceptualizations
After generative AI became publicly available, AI literacy research increasingly addressed tool use, prompt engineering, critical evaluation, and indirect benefits. These studies expanded across higher education and K-12 curricula while leaving uneven empirical coverage.
- Twenty-two studies, or 18%, adopted only the AI-tool perspective, generally emphasizing effective use of generative AI tools.
- Sixteen studies in 2024 by July addressed generative-AI tool use, while ChatGPT was frequently treated as the main tool students should master.
- Seven studies isolated prompt engineering as a specific AI-literacy competency and developed activities or assessments for practicing it.
- Researchers increasingly combined functional and critical literacy by teaching effective tool use alongside ethical, societal, and labor-market implications.
- Indirect-benefit studies emphasized outcomes such as STEM interest, computational thinking, creative thinking, or more positive attitudes toward AI.
5 Discussion
The review organizes AI literacy studies through three AI perspectives and three literacy perspectives, then uses the framework to interpret post-generative-AI trends. It identifies terminology needs and gaps in empirical and cross-perspective research.
- The framework classifies AI as technical details, tools, or sociotechnical systems, and literacy as functional, critical, or indirectly beneficial.
- The framework explains definitional ambiguity because different AI and literacy perspectives suit different learning contexts and objectives.
- The review calls for more specific terminology because “AI literacy” covers skills requiring substantially different curricular implementations.
- The review identifies gaps in combining AI tools with critical or sociocultural perspectives and in combining tools with technical AI details.
- Since generative AI, research shifted toward post-secondary AI-tool use, while K-12 curricula incorporated tools through mediated activities.
- The review also identifies limited empirical post-secondary research and a need to translate comprehensive K-12 curricula into post-secondary contexts.
6 Limitations
The review’s framework maps AI literacy conceptualizations but does not evaluate intervention effectiveness and excludes several education settings. Its generalizability therefore requires future verification.
- The integrative review did not address specific research questions within AI literacy.
- It did not evaluate learning-intervention effectiveness because the interventions differed in research questions and methodologies and were not directly comparable.
- The frequency of AI-literacy conceptualizations should not be interpreted as evidence of their effectiveness.
- The review excludes pre-service teacher, adult, healthcare, nursing, and business AI literacy settings.
- Future work must verify whether the conceptual framework generalizes to the excluded contexts.
7 Conclusion
The review frames AI literacy as connected but distinct competencies across three literacy perspectives and three AI perspectives. It identifies major post-generative-AI shifts, supports more specific terminology, and points to research gaps in critical literacy and comparable post-secondary curricula.
- Framework: The framework combines functional, critical, and indirectly beneficial literacy with technical-detail, tool, and sociocultural perspectives on AI.It is based on a review of 124 studies in K-12 and post-secondary contexts.
- Research shifts: Post-generative-AI research shifted toward post-secondary studies emphasizing effective use of generative AI tools, including prompt engineering.K-12 research also incorporated more AI-tool elements.
- Research shifts: Studies increasingly addressed indirect benefits of AI literacy, including increased interest in STEM.This represents an additional literacy perspective beyond direct technical or tool-use objectives.
- Implications: The generic label “AI literacy” covers substantially different interventions, motivating more specialized terms based on distinct AI and literacy conceptualizations.Greater specificity is intended to help researchers and practitioners communicate objectives and approaches more clearly.
- Research gaps: The review identifies research gaps in promoting critical literacy among students interacting with AI tools.It also notes a gap in developing comparable AI literacy curricula for post-secondary settings.