Writing Ethically With AI: How The Writing Competence Of Students And Teachers Is Evolving

The widespread adoption of Generative Artificial Intelligence (GAI) in educational and academic contexts has fundamentally altered the conditions under which writing takes place. Neurocognitive research has long established that writing is a process in which movement, memory, and reasoning are deeply intertwined, and recent empirical evidence suggests that delegating core compositional functions to AI models may generate measurable cognitive deficits over time. The cognitive work that GAI tends to remove is not incidental to academic writing: it is constitutive of it. The critical issue is therefore not the use of GAI itself, but the extent to which such use preserves or progressively redistributes the cognitive processes that writing is meant to develop. This paper addresses this problem through a systematic screening of 62 recent scientific contributions, from which 20 key tools have been identified and categorized according to their capacity to support, redistribute, or replace core cognitive functions involved in academic writing. The tools are organized into three functional categories, source research and discovery, literature analysis and synthesis, and linguistic and stylistic editing, with Large Language Models treated as a distinct class, given their cross-cutting generative role and the particular cognitive stakes of prompt construction and dialogue management. Based on this taxonomy, the paper proposes an integrated usage strategy that sequences the selected tools across the writing process, explicitly aligning their use with the preservation of the author's intellectual control. Drawing on Jevons' paradox and recent neuroscientific findings, the discussion argues that GAI does not erode writing competence uniformly but redistributes it, with a structural risk of cognitive weakening wherever intellectual agency is not actively preserved. Writing competence is therefore not declining but being reconfigured: the challenge is not primarilytechnological but pedagogical, requiring students and teachers to use powerful tools without delegating what only human thought can do.