Learning Behavior & AI Agency
Workshop on Predicting Performance from Reading and Learning Behavior and Orchestrating Human and AI Agency — Joint Edition
Workshop on Predicting Performance from Reading and Learning Behavior and Orchestrating Human and AI Agency — Joint Edition
This ICCE 2026 edition brings two established workshops together. Digital learning has made rich behavioral data available, while educational AI is shifting from reactive tools toward proactive, agentic systems. We unite both around one design question: how can agentic AI use behavioral data to deliver proactive interventions while preserving learner agency and teacher professional judgement?
Over the past decade, Learning Analytics has established robust methods for analyzing digital learning traces — reading behavior, navigation, duration, and annotation patterns from digital textbooks and e-book systems — supporting at-risk identification, performance prediction, and evidence-based instructional interventions. Concurrently, Large Language Models and Generative AI are transforming educational AI from reactive, prompt-based tools into proactive Agentic AI systems that act as active collaborators rather than passive tools.
Yet performance gains do not guarantee learning: AI can improve task outcomes without producing meaningful learning unless guided by pedagogical principles. This defines a new design challenge at the intersection of AI autonomy and pedagogical alignment, and motivates deeper integration of learning analytics into research on human-AI interaction — from structured learning logs to dynamic interaction data such as conversational processes, prompt refinement, help-seeking, and collaboration patterns.
Aligned with the ICCE 2026 theme "Reimagining Learning Ecologies in the Age of Intelligent Technologies", this half-day workshop bridges ICCE's sub-conferences on AIED/ITS (C1) and Advanced Learning Technologies and Learning Analytics (C3), convening researchers across learning analytics, AI in education, and human-AI interaction.
We welcome full papers, short papers, and extended summaries on (but not limited to) the following topics:
All dates are Anywhere on Earth (AOE).
All submissions must follow the ICCE 2026 paper template.
Comprehensive studies with complete methodology and findings
Work-in-progress with preliminary results or novel concepts
Perspectives, provocations, or emerging directions
Review process: All submissions undergo single-blind review by at least two program committee members.
Submission portal: Submit via EasyChair.
Template: ICCE 2026 paper template (Word).
Proceedings: Accepted papers will appear in one volume of the workshop proceedings with ISBN and will be indexed by Elsevier Bibliographic Database. Published workshop papers will be made available on the official ICCE 2026 website.
Since 2018, data challenge workshops held alongside ICCE and LAK have given researchers something still rare in our field: access to real learning data for a shared research problem. In this keynote, I reflect on what the series has taught us about anonymisation, data agreements, and the challenges of deciding what data can be shared, with whom, and under what conditions.
I also draw on a recent survey of 1,125 papers from LAK, EDM, and AIED, which identified 172 open datasets used in 204 publications and informed the PRACTICE guidelines for publishing educational data. Finally, I discuss how learning data are changing as chatbots and agentic AI enter learning environments, shifting the record of learning from clickstreams to conversations and creating new challenges for data sharing and privacy.
When does AI help students think, and when does it do the thinking for them? Drawing on our research integrating AI and HCI, this keynote explores how learners’ intentions and self-regulation shape their interactions with chatbots, and how interaction design can encourage deeper thinking. I connect these studies to a central challenge for proactive agentic AI: deciding when and how to intervene while preserving learner agency. I outline a design direction in which AI interprets engagement through ongoing interaction, offers appropriate support, and uses learners’ responses to guide its next move. The goal is AI that helps students question, explain, and develop their own ideas.











