Abstract:

Recent scholarship in Artificial Intelligence in Education (AIEd) has increasingly conceptualized artificial intelligence as a learning partner, tutor, collaborator, or pedagogical agent. Although such perspectives illuminate the interactive potential of intelligent systems, they frequently remain functionally oriented and may imply forms of symmetrical agency between learners and technology that are theoretically difficult to reconcile with sociocultural understandings of mediation and multiliteracies conceptions of literacy as socially situated, dialogic, and multimodal practice. Within these dominant framings, Artificial Intelligence (AI) is often treated primarily through its operational affordances, while insufficient attention is given to the pedagogical conditions under which educational meaning is constructed, negotiated, and transformed.

In response, this paper introduces AI as a literacy companion as a new conceptual framework for interpreting AI-supported literacy learning in primary education. Rather than positioning AI as an equal pedagogical partner, the concept frames AI as a pedagogically bounded presence whose educational significance lies in accompaniment, scaffolding, dialogic support, and multimodal expansion of meaning-making while preserving human agency, authorship, interpretive responsibility, and teacher mediation. The framework seeks to bridge three domains that are rarely theorized in relation to one another: AI learning companion systems, sociocultural theories of mediated learning, and multiliteracies approaches to literacy development.

To operationalize this conceptual contribution, the study proposes three pedagogical modes of AI integration: AI as design support, where AI contributes to instructional preparation and pedagogical structuring; AI as bounded scaffold, where AI supports comprehension monitoring, feedback, and guided participation within clearly delimited learning tasks; and AI as reflective co-design space, where AI functions as a dynamic environment for collaborative redesign, multimodal production, and reflective meaning-making. These modes are differentiated by their pedagogical function, the type of learner engagement they activate, and the additional educational value they generate when embedded through structured and reflective use.

The framework is informed by evidence from a design-based classroom intervention with fourth-grade students, in which AI supported reading comprehension through asynchronous formative feedback and later served as a generative visual scaffold for collaborative narrative transformation. Findings suggest that when pedagogically orchestrated, AI extends literacy learning not through automation, but through critical and creative participation in evolving multimodal literacy practices.

The paper argues that the educational value of AI emerges not from technological autonomy but from instructional design, pedagogical intentionality, and the quality of mediated participation. By proposing AI as a literacy companion, the study offers an original analytic lens for human-centred AI integration and contributes to contemporary debates on responsible educational innovation, teacher design capacity, and the future of literacy pedagogy.

Keywords:

Artificial Intelligence in Education, Multiliteracies, Sociocultural Learning, Instructional Design, Primary Education.