<![CDATA[Generative artificial intelligence is increasingly being used to design learning plans, generate learning materials, support learning interactions, and provide feedback. These developments not only raise questions regarding the effectiveness of the technology, but also alter the distribution of authority in pedagogical decision-making. This article aims to analyse how pedagogical control is distributed between teachers and generative artificial intelligence during the planning, implementation, and assessment stages of learning. The research employs a critical-integrative literature review of 28 substantive publications from the period 2023–2026. Two methodological frameworks were used to guide the synthesis process, whilst one policy document was utilised to strengthen the analysis of teachers’ agency. The literature data were analysed through thematic coding, cross-source comparison, and critical interpretation of the technology’s functions, teachers’ authority, and the risks of delegating decision-making. The findings indicate that generative artificial intelligence is best suited to carrying out limited and revisable operational control, such as generating design alternatives, modifying materials, simulating interactions, and compiling initial feedback. Conversely, teachers must retain control over strategic, contextual, relational, evaluative and ethical aspects. Uncontrolled delegation may result in curricular misalignment, factual errors, homogenisation of learning, a reduction in pedagogical relationships and a blurring of accountability. This article proposes the concept of ‘asymmetric distribution of pedagogical control’, namely a division of functions that utilises the generative capacity of AI without equating the machine’s authority with the teacher’s professional responsibility. The findings confirm that the quality of AI integration is not determined by the extent of automation, but rather by the clarity of delegation boundaries, validation mechanisms, and the sustainability of teacher agency.]]>
