MAKING REASONING VISIBLE: PORTFOLIO-BASED ASSESSMENT AND AI-USE DISCLOSURE IN EFL ACADEMIC WRITING

Generative artificial intelligence (GenAI) has weakened the link between a well-written essay and the learning that is supposed to stand behind it. This thesis argues that, in EFL academic writing, the problem is primarily one of assessment rather than of technology. If only the final product is graded, students rationally choose the shortest route to a polished text. The paper proposes a process-oriented assessment model that combines a writing portfolio, a six-criterion rubric for critical analytical writing and a mandatory AI-use disclosure. The model makes the learner's reasoning observable at each stage and turns responsible GenAI use into an assessable learning outcome. Its application in ESP and General English classes in Uzbekistan is discussed, together with directions for empirical validation.