Preserving human agency amid AI-mediated thinking and learning.
Critical AI Literacy (CAIL) is a framework for preserving human agency amid AI-mediated thinking and learning. Drawing on Critical Literacy Studies, the New Literacies Movement, and Information Literacy, CAIL extends Paulo Freire's call to "read the word and the world" to also read the machine.
"Literacy concerns what counts as meaning: it is embedded in social and cultural practices, involves the negotiation between symbols and action, and is mediated by cognitive artifacts that encode power relations."
Complete avoidance of AI interaction has become increasingly untenable. Most current AI literacy frameworks only address tool mastery or ethical awareness. However, when literacy practices produce claims about the world, they generate knowledge. As a cognitive artifact, GenAI distorts this process in ways that demand critical engagement.
Generative AI produces outputs that arrive as coherent knowledge (fluent, confident, apparently authoritative) yet they lack epistemic requirements: grounding, perspective, consequence, intent, or understanding.
This gap between appearing to know (fluency) and actually knowing (understanding) is invisible to most users, concealed by the very fluency that makes GenAI systems useful.
CAIL trains people to recognize and interrupt this naturalization.
As a "denaturalization posture," CAIL requires a set of interdependent dispositions of discernment that deepen through practice.
Use and understand GenAI without surrendering agency. Master the tools while maintaining critical distance from their outputs and processes.
Interrogate operations, outputs, and training data. Detect fabrications, surface bias, and recognize ulterior motives embedded in AI systems.
Examine how AI reorients cognition, values, and practices. Understand how your thinking changes through repeated AI interaction.
Understand how AI can both widen access and deepen inequity, extraction, power, or surveillance. Recognize how AI reshapes communities of practice and collective agency.
Interrogate the frames through which AI naturalizes interpretations of reality. Examine how meaning is constructed and presented as neutral.
These questions operationalize the five core CAIL capacities by translating the framework into a protocol of critical inquiry: making provenance traceable, boundaries visible, consequences accountable, power legible, and conceptual frames explicit.
The aim is to develop disciplined discernment: an ongoing practice of noticing how meaning is constructed, how it travels, and how easily coherence can be mistaken for understanding.
Fluency masking ignorance, subtle errors, unjustified confidence
Neutrality claims, stereotype leakage, opaque business motives
Every AI interaction trains you; attend to what you're learning.
Dependency formation, skill atrophy, diminished critical thinking
Concentration of authority, invisible gatekeeping, naturalized hierarchies
Interrogate the frames through which AI naturalizes interpretations of reality.
Neutrality claims, stereotype leakage, opaque business motives
Isolation from peers, declining collaborative work, loss of shared standards, erosion of collective expertise
Access the archive, contribute your experiences, and join the conversation.
A curated collection of resources, cases, and examples for developing critical AI literacy in practice.
Visit ArchiveShare your examples of AI outputs that demand critical analysis—help build our collective understanding.
ContributeAccess the original Critical AI Literacy framework document with full citations and references.
View Document