Overview
What the framework is trying to do
Zhang and colleagues frame Critical AI Literacy as responsible, critical interaction with AI rather than a competition between human and machine intelligence. Their model foregrounds data practices, data sovereignty, digital colonialism, labor, environmental impact, automation bias, and digital ableism.
RACBAC provides the practical evaluation layer: Relevance, Accuracy, Coverage, Bias, Authority, and Currency. This makes the framework especially useful where system critique must connect directly to research verification.
Framework structure
Core dimensions or competencies
Critical system awareness
Data transparency, sovereignty, environmental impacts, labor, automation bias, and digital ableism.
Human-AI interaction
Responsible hybrid intelligence and judgment about when, why, and how much AI should be used.
RACBAC
Relevance, Accuracy, Coverage, Bias, Authority, and Currency as an evaluation structure.
Critical depth
Why this framework is rated Very High
Rated Very High because it explicitly connects structural concerns such as data sovereignty, labor, environment, ableism, bias, and colonialism with practical verification and human judgment.
Central
Bias, labor, equity, accessibility, environmental impact, epistemic authority, verification, transparency, human agency
Present
Power, privacy, surveillance
Limited
None among the dimensions currently indexed
Not evident
See the Critical Dimensions Index for dimensions not assigned a stronger classification.
Evidence base
What supports the framework?
Scholarly synthesis and applied evaluation framework
Practical use
Especially useful for
Research instruction, library instruction, verification workflows, AI-assisted research, source evaluation, and critical information literacy.
Evaluation
Strengths and limitations
Strengths
- Bridges critical systems analysis and hands-on research verification.
- Strong relevance to library and information literacy practice.
- Explicitly includes disability and data sovereignty.
Limitations
- Applied framework is newer and has a smaller evidence base than long-established information-evaluation models.
- Some dimensions are synthesized from adjacent literature rather than empirically validated as a single scale.
Portability
How well does it travel?
High. The evaluation components and critical dimensions are adaptable across disciplines and institutional types.
Source
Citation
Zhang, et al. (2025). Critical AI literacy and RACBAC. International Journal of Librarianship, 10(2). https://doi.org/10.23974/ijol.2025.vol10.2.431