Evidence Map & Research Gaps
This map summarizes patterns in the current project corpus and distinguishes areas with substantial descriptive evidence from areas where intervention studies, longitudinal evidence, validated measures, or diverse institutional contexts remain limited.
Strong indicates a substantial and recurring body of literature in the current corpus. Moderate indicates meaningful evidence with important limitations. Emerging indicates a growing area with relatively few studies or uneven methods. Sparse indicates a clear research opportunity. These labels describe the reviewed corpus and should be updated as the literature grows.
A large and growing body of surveys and qualitative studies examines how students use GenAI, what they value, and what concerns them. Accuracy, immediacy, privacy, integrity, confidence, and policy uncertainty recur across this work.
Academic integrity, privacy, bias, responsible use, and over-reliance appear frequently across empirical studies, reviews, and competency frameworks.
The field now contains multiple AI-literacy and critical-AI-literacy frameworks spanning students, teachers, institutions, libraries, and critical pedagogy. The challenge is increasingly comparison and validation, not simple absence.
Teacher and faculty frameworks are well developed, and empirical studies document uneven adoption and confidence. Evidence is thinner on which professional-learning interventions produce durable changes in teaching practice.
Policy clarity, assessment redesign, transparency, and governance are recurring priorities, but institutional approaches remain uneven and are often described rather than evaluated longitudinally.
Studies consistently identify weak evaluation of AI output and the need for scaffolds. Intervention examples are increasing, but validated measures of real-world verification behavior remain limited.
Critical AI literacy work increasingly addresses power, surveillance, justice, representation, and action. However, many interventions remain small-scale, conceptual, or concentrated in youth contexts.
Academic-library research documents staff readiness and a growing number of instructional interventions. Evidence of library-led effects on student learning is still comparatively limited.
Accessibility is prominent in some critical frameworks, especially OU and Zhang, but empirical higher-education studies that center disabled learners remain scarce.
These dimensions are increasingly visible in critical frameworks but remain far less common in empirical AI-literacy interventions than bias, privacy, or academic integrity.
Community colleges are underrepresented in published AI-literacy research relative to universities, despite distinct student populations, teaching loads, access issues, and support structures.
There is growing discussion of AI-assisted research, but comparatively little empirical work isolates doctoral learners' critical-AI-literacy needs across literature review, data analysis, authorship, and research integrity.
Most studies capture perceptions, intentions, or short-term experiences. Evidence showing how critical AI literacy develops over semesters or years remains thin.
Many frameworks define competencies or dimensions, but few measures capture authentic critical judgment across verification, power, equity, agency, and context rather than self-reported confidence.
Frameworks are often presented as transferable, but relatively few studies directly test the same model across community colleges, research universities, doctoral programs, or national contexts.
Professional staff, support staff, and administrators are much less studied than students and faculty, even though AI policies and services depend on their work.
Research agenda
What supports work for students and instructors in open-access, teaching-intensive institutions with different resources, student demographics, and support structures?
Can learners actually verify claims, identify missing perspectives, make transparent AI-use decisions, and explain when not to use AI?
Which instructional approaches produce measurable gains in critical AI literacy rather than short-term satisfaction or self-reported confidence?
How does critical AI literacy change as learners gain disciplinary expertise and repeated experience with AI?
What combinations of policy, professional development, instructional design, library partnership, and peer support lead to durable changes in teaching?
How can instruction meaningfully address labor, environment, surveillance, disability, data sovereignty, epistemic authority, and corporate power without turning them into token checklist items?
What does critical AI literacy look like in literature searching, synthesis, data work, coding, authorship, peer review, and research integrity?
Which frameworks travel successfully across institution types, disciplines, countries, and learner populations, and what must change when they do?
The larger pattern
Student perceptions, adoption, ethics concerns, and attitudes are increasingly well documented. The harder questions now concern development and transfer: whether instruction changes verification behavior, whether critical judgment persists over time, how faculty support affects student practice, and whether frameworks designed in one institution or country work in another.
The critical side of AI literacy is also uneven. Bias, privacy, and academic integrity are common. Labor, environmental impact, surveillance, disability, data sovereignty, epistemic authority, and corporate power appear much less consistently, especially in empirical intervention studies.