The proliferation of generative artificial intelligence (GenAI) in educational contexts has predominantly been conceptualized as an instrumental enhancement to existing pedagogical frameworks, thereby neglecting its profound transformative capacity. This reductionist perspective fails to recognize GenAI's potential as a cognitive collaborator within the learning ecosystem. The genesis of this research emerges from our theoretical proposition that the actualization of GenAI's educational potential necessitates the development of specialized metacognitive competencies—which we term Meta-AI [1] skills—that enable learners to engage in substantive epistemic partnerships with artificial intelligence systems.

This study explores the multifaceted transformative dimensions of GenAI that necessitate the development of Meta-AI skills among learners and educators. Our analysis identifies several critical shifts occurring within educational contexts. First, we observe a transition from symbolic to indexical information processing frameworks, wherein the production of dynamic, multimodal outputs by AI systems requires increasingly sophisticated interpretive competencies. Second, we argue that traditional epistemological approaches grounded in scientific generalization are being displaced by methodologies that privilege situated reasoning and adaptive exploration within specific learning contexts. Third, the inherent multimodal capabilities of contemporary GenAI systems function as catalysts for enhanced metacognitive awareness. Through empirical investigation, we demonstrate how AI's capacity to synthesize diverse representational modalities—encompassing textual, visual, auditory, and video formats—facilitates the development of advanced cognitive flexibility and self-reflective practices among learners. These interactive dynamics fundamentally challenge conventional conceptualizations of artificial intelligence as an epistemologically opaque "black box," instead positioning students as active epistemic agents in their cognitive development rather than passive recipients of AI-generated content.

This research employs a threshold concepts pedagogical framework [2] to facilitate the acquisition of essential metacognitive competencies among learners. We conceptualize the mentioned shifts as critical thresholds that necessitate metacognitive perception for effective traversal. Accordingly, pertinent teacher education curricula have been reconceptualized to center these threshold concepts as organizing principles.

The empirical component of this study presents qualitative findings derived from semi-structured interviews with teacher-participants who engaged with the redesigned professional development program. Analysis of these data indicates that the pedagogical approach articulated herein effectively catalyzes the development of metacognitive capabilities among the majority of participating students. Furthermore, our findings demonstrate significant transformative outcomes in students' conceptual mastery of complex phenomena, suggesting that threshold-based Meta-AI skill development represents a promising direction for educational innovation in the era of AI integration.

References:
[1]⁠ ⁠Sidorkin, A. M. (2025). AI Integration Blueprint: Transforming Higher Education for the Age of Intelligence. AI-EDU Arxiv.
[2] Meyer, J., & Land, R. (2003). Threshold concepts and troublesome knowledge: Linkages to ways of thinking and practising within the disciplines.

Keywords: Digital technologies, artificial intelligence, generative AI, education, metacognitive skills.