Sİ DƏSTƏKLI FORMATIV QIYMƏTLƏNDIRMƏ: RUBRIKALARA ƏSASLANAN GENERATOR Sİ VASITƏSILƏ AKADEMIK YAZIDA RƏYIN SƏMƏRƏLILIYININ VƏ KEYFIYYƏT DƏRINLIYININ OPTIMALLAŞDIRILMASI

Abstract

In higher education, academic writing serves not only as a tool for assessing knowledge but also as a fundamental cognitive process that develops students' critical thinking, reasoning, and scholarly communication skills. The development of these skills heavily relies on the quality, frequency, and content of formative feedback. However, a major challenge modern higher education institutions face is resource and time constraints, particularly in large classes where providing detailed, personalized feedback on every essay or paper demands substantial time and psychological energy from instructors. This often forces educators to choose between delaying feedback for weeks or resorting to generic, superficial comments. The integration of Generative Artificial Intelligence (GenAI) technologies, specifically Large Language Models (LLMs), offers a promising solution to this dilemma by analyzing logical structures, depth of argumentation, and stylistic features to generate high-level natural language feedback. This article explores the theoretical and practical foundations of a rubric-based, AI-supported formative assessment model while maintaining Human-in-the-Loop (HITL) oversight. By examining the pedagogical theories of formative feedback, the evolution of automated assessment systems, and a structured three-stage institutional implementation roadmap, this study demonstrates how the proposed model can reduce grading workloads, enhance feedback depth, and preserve pedagogical subjectivity while mitigating ethical and technical risks such as algorithmic bias and data privacy concerns.