PEDAGOGICAL POTENTIAL AND LIMITATIONS OF AI TOOLS FOR ACADEMIC WRITING INSTRUCTION

Abstract

The increasing use of artificial intelligence tools in higher education has created new opportunities for academic writing instruction while simultaneously raising concerns about learner dependence, authorship, academic integrity, source reliability, and the development of independent thinking. This article examines the pedagogical potential and limitations of AI tools in academic writing instruction from a theoretical and analytical perspective. Particular attention is paid to the use of generative AI for brainstorming, outlining, language support, formative feedback, revision, vocabulary development, and individualized learning. The analysis also considers major limitations, including hallucinated information, fabricated references, overreliance on automated suggestions, homogenization of writing style, reduced cognitive engagement, and difficulties in assessing authentic student performance. The study argues that AI tools are most effective when they function as supplementary pedagogical resources rather than substitutes for student reasoning, source evaluation, and authorship. Effective integration requires clear instructional objectives, process-oriented assessment, AI literacy, transparent disclosure practices, teacher mediation, and systematic verification of AI-generated information. The article concludes that the educational value of AI in academic writing depends less on the technological capabilities of the tools themselves than on the pedagogical framework within which they are used.