Rethinking Generative AI Literacy: An Integrative, Developmental, and Dialectical Framework for K–12 Teacher Education

Generative AI has integrated into classrooms at a pace that far exceeds the development of teacher preparation programs. This creates a 'literacy lag,' where educators are expected to use tools like ChatGPT or Claude without the necessary conceptual, pedagogical, or ethical training. Existing AI literacy frameworks, largely developed for earlier, rule-based machine learning systems, fail to address the unique challenges of generative models, such as hallucination, prompt dependency, and the erosion of cognitive processes through uncritical offloading.

To address this gap, the authors developed the Responsible AI Literacy in Education (RAIL-Ed) framework through a systematic review of 67 studies. The framework is built upon four foundational traditions: Freire’s critical pedagogy (agency), Dewey’s pragmatism (reflective inquiry), Vygotsky’s sociocultural theory (mediated cognition), and Shneiderman’s human-centered design (high automation with high human control).

RAIL-Ed consists of six interdependent pillars:

1. Technical Fluency: Understanding how generative models function.
2. Critical Evaluation: Assessing the accuracy and provenance of outputs.
3. Human-AI Collaboration: Engaging with AI as a mediational agent.
4. Contextual Awareness: Recognizing the cultural and linguistic settings of AI use.
5. Ethical Reasoning: Navigating bias, copyright, and epistemic risks.
6. Empowered Agency: Exercising the authority to adopt, restrain, or refuse AI usage.

RAIL-Ed distinguishes itself by being integrative, developmental, and dialectical. It posits that the absence of any single pillar leads to pedagogical failure. By treating equity and ethics as constitutive rather than optional, the framework provides a robust basis for teacher education that prepares educators to navigate the complex, evolving landscape of generative AI without ceding their pedagogical authority. It offers a roadmap for moving beyond simple 'prompt engineering' toward a deeper, critical engagement with AI as a transformative educational tool.