From Prompt to Prose: What an AI Writing Tutor Reveals About Student Learning

Abstract

The proliferation of GenAI presents both profound opportunities and significant challenges for higher education, particularly in teaching foundational skills within online learning environments. This presentation examines the implementation of Lexi LawScribe, a bespoke AI writing tutor, in a first-year law unit. Adopting the GenAI-TPACK framework (Belkina et al, 2025), we analyse the integration of Lexi within the pedagogical requirements of legal writing instruction and the content demands of criminal law. The unique online delivery of the course provides a critical contextual knowledge lens, highlighting the need for scalable, on-demand student support. This presentation draws on an analysis of student usage logs from the AI tutor. We explore the patterns of student interaction to understand how they leverage the tool for learning. Initial findings indicate that students primarily use the AI tutor for scaffolding complex legal writing structures, receiving iterative feedback to refine their analysis, and reducing the cognitive load associated with mastering discipline-specific communication conventions. We conclude by providing practical, evidence-based recommendations for the future design of AI-driven learning support, aiming to enhance student engagement and success in online and blended learning contexts.