Abstract:

As artificial intelligence becomes increasingly embedded in K–12 learning environments, many systems claim the ability to identify what students know and to prescribe next steps for instruction. This paper argues that such automated approaches risk displacing the very human processes that make learning possible, including reasoning, discourse, identity development, and collaborative sense-making. Drawing on the NCSM Educational Technology and AI Guidance for Math Leaders and analyses of teacher interactions with a purpose-trained GPT, the paper examines how AI’s value shifts when educators use it not as a surrogate for instruction but as a partner for contextualizing learning, enriching cultural and place-based relevance, and supporting professional inquiry.

Findings from workshop-based practitioner research demonstrate that educators’ critiques, revisions, and collective questioning of AI outputs reveal a form of AI literacy grounded in teacher agency rather than tool compliance. When teachers work in pairs or triads to interrogate AI responses, they preserve the social and dialogic dimensions of learning that automated systems often overshadow. These interactions highlight the centrality of professional judgment in evaluating the appropriateness, accuracy, and cultural responsiveness of AI-generated content.

The analysis proposes a conceptual shift from AI as a mechanism for instructional automation to AI as an amplifier of human thinking. This reframing positions educators—not algorithms—as the primary interpreters of student learning and as co-constructors of technological use. The paper concludes by outlining implications for teacher preparation, leadership decision-making, and ethical technology adoption, arguing that AI-integrated learning environments must foreground human presence, inquiry, and identity if they are to support meaningful and equitable learning.

Keywords:

Artificial intelligence in education, AI literacy, Teacher professional judgment, Student thinking, Human centered learning.