This article analyzes the challenges of teacher education in relation to artificial intelligence, with attention to students' productive inclusion. The objective was to understand which professional forms of knowledge help teachers integrate generative systems into teaching without surrendering pedagogical autonomy, student authorship, or data protection. The study adopted a narrative literature review and documentary analysis of fifteen open sources published between 2021 and 2026. The corpus included Brazilian articles, documents from the Ministry of Education, national legislation, and international frameworks on artificial intelligence competencies and work. The analysis identified four educational demands: technical literacy, pedagogical decision-making, ethical responsibility, and interpretation of occupational change. The results show that operational command of tools is insufficient. Teachers need to verify outputs, recognize bias, redesign assessment, protect personal information, and relate school content to professional tasks mediated by intelligent systems. Productive inclusion requires equitable access, continuing education linked to practice, and teacher participation in institutional rules. Artificial intelligence gains educational value when it remains subordinate to learning goals and human judgment. Keywords: Artificial intelligence. Teacher education. Productive inclusion