LDFacilitator: A System for Facilitating Learning Design Using Large Language Models

Learning design is the systematic process of articulating learning outcomes, selecting and sequencing activities, and aligning assessment to ensure teaching intentions are translated into effective student experiences. Learning designs grounded on robust pedagogical principles is a pre-requisite for effective teaching. However, pre-service teachers often struggle with challenges such as navigating digital learning design on learning management systems such as Moodle. Learning design systems grounded on robust learning sciences research, such as the Learning Design StudioTM, while feature-rich, require considerable conceptual maturity and technological proficiency to construct clear, measurable learning outcomes, learning sequences, and assessments that are pedagogically aligned. As a result, many novice educators underutilize the capabilities of learning design platforms and produce learning designs that lack precision and effectiveness.

Recent advances in large language models (LLMs) offer a promising solution. Models such as those in the GPT series can generate tailored explanations and deliver timely feedback based on human requests. For novices, the conversational affordances of LLMs are particularly valuable: rather than relying on static tutorials, users can ask contextualized questions (e.g., “Is this outcome measurable?”), receive immediate, relevant feedback, and refine their work directly within the same environment. This interactivity enables the embedding of “just-in-time” expertise within learning design systems.

To operationalize this opportunity, we present LDFacilitator, an LLM-powered assistant fully integrated into the Learning Design StudioTM (LDS) platform. The LDS platform is an interactive, web-based environment developed to support teachers in designing, visualizing, and implementing structured instructional sequences. Within LDS, LDFacilitator serves two core functions: (1) it orients users by providing conversational guidance for navigating LDS features, ensuring teachers can efficiently access tools relevant to each stage of the design process; (2) it offers real-time, rubric-aligned feedback on critical learning design components, drawing on established pedagogical-design principles. For example, when users draft a learning outcome, LDFacilitator analyzes the statement, highlights its strengths, identifies any ambiguities, and suggests revisions to enhance clarity and measurability. This dual support aims to lower the barrier to effective platform use and scaffolds the development of pedagogically sound learning design artefacts.

An openly accessible demonstration (https://youtu.be/hL7fMruW7PY) showcases LDFacilitator’s capabilities. By integrating advanced LLMs within the LDS platform, LDFacilitator exemplifies the potential of human–AI collaboration to democratize learning design expertise, strengthen teacher preparation, and enable scalable, data-informed professional development across varied educational contexts.