Abstract

Title: Rethinking Learning Analytics for the Generative AI Era

Learning analytics is a mature field dedicated to harnessing digital data to better understand and improve learning and teaching. The emergence of generative artificial intelligence (GenAI) introduces new questions—and possibilities—for how we analyze and shape educational experiences. This keynote will examine the evolving relationship between GenAI and learning analytics, focusing on two key directions. First, it will consider how GenAI is reshaping the educational landscape, creating contexts where learning analytics can play a critical role in evaluating GenAI’s effectiveness and guiding its responsible use. Second, the talk will explore how GenAI can itself become a driver of innovation in learning analytics, offering new capabilities for data generation, interpretation, and feedback. Drawing on findings from a series of empirical studies, the keynote will highlight both conceptual and practical implications for educators, researchers, and technology developers navigating this rapidly changing terrain.
 

Title: Transforming Education with LLM-Driven Learning Analytics

Integrating Large Language Models (LLMs) into Learning Analytics (LA) reports presents significant opportunities for enhancing teaching and learning. This presentation examines the types of data most beneficial for educators and how Learning Analytics Platforms (LAP) can facilitate evidence-based improvements in educational practices. By analysing both quantitative and qualitative data, LLMs provide actionable insights that educators can use to refine their teaching strategies. Addressing the challenge of hallucination in LLM outputs, the Chain-of-Verification (CoV) technique is introduced as a method to ensure accuracy and reliability. Additionally, best practices in prompt engineering are shared to optimise LLM outputs, making them more usable and relevant for educational contexts. This integration of LLMs into LA reports aims to drive meaningful enhancements in educational outcomes.