Skip to content

Publications

2026a: μEd API: Towards A Shared API for EdTech Microservices

2026b: Rethinking Scaffolding in LLM Tutors: The Interactional Mismatch Between Benchmarks and Real-World Deployments

  • Download PDF or access via arxiv: https://arxiv.org/abs/2606.15766

  • Conference paper, presented at the Pluralistic Alignment Workshop @ ICML 2026; Observed that students in real-world deployments take up the scaffolding provided by LLM chatbots far less than what AI tutor benchmarks assume, highlighting a possible mismatch between a chatbot’s pedagogical framing and the student’s learning goals.

2026c: “How Do I ... ?”: Procedural Questions Predominate Student-LLM Chatbot Conversations

  • Download PDF or access via arxiv: https://arxiv.org/abs/2602.18372

  • Conference paper; Analysed student messages from two STEM learning contexts (formative self-study and summative coursework), classifying student questions with LLM and human raters across existing schemas. Discusses the limits of schema-based classification of student questions to LLM chatbots.

2026d: Evaluating LLM-Generated Formative Feedback for Undergraduate Mathematics Through the Lens of Feedback Theory

2025a: Formative feedback on engineering self-study: Towards 1 million times per year per cohort

2025b: Exploring the feasibility of using Generative AI to emulate teacher feedback for GCSE English Language assessments

2025c: How do we define and evaluate "good" automated formative feedback?

  • Download PDF. A workshop hosted at SEFI 2025.

  • General overview (6 pages) of the literature and thinking behind automated formative feedback, with some community inputs on how we should evaluate automated feedback. Good for educators wanting a general framing of our approach.

2025d: AI microservices for sustainable innovation in education

  • Download PDF pre-print under review. Also available on OSF: https://osf.io/preprints/edarxiv/wq4bd_v3

  • Journal pre-print (~30 pages). Extended manifesto on why the education sector should use microservices for automated judgement in education. Literature review, conceptual arguments on innovation and ethics, and results from deployments of Lambda Feedback. This paper is a long read, providing the detailed thinking behind Lambda Feedback.

2025e: Chatbots for Dialogic Feedback during Self-Study: The Importance of Contextual Information

  • Download PDF or access via the Engineering Education Research Network (EERN) Conference Proceedings

  • Conference Extended Abstract (8-pages). Using a framework of impasse-driven learning we explore the role of chatbots, and focus on whether the bot has access to context. Empirical results show students benefitting from chatbots accessing context. Valuable thinking behind the benefits and risks of AI in education, combining theory and practice.

2024: Automated Feedback on Student Attempts to Produce a Set of Dimensionless Power Products from a Set of Physical Quantities that Describe a Physical Problem

Blog articles