A Human‐in‐the‐Loop Framework for AI‐Generated, Accreditation‐Aligned Assessments in Engineering Education

ABSTRACT

The integration of Generative Artificial Intelligence (GenAI) into engineering education offers significant potential for assessment creation, yet generic Large Language Models (LLMs) often fail to adhere to the strict technical and pedagogical constraints required for engineering programs. Specifically, standard prompting approaches struggle to generate valid Moodle XML syntax, render complex mathematical notation (LaTeX) without errors, and align assessments with granular accreditation standards such as those defined by ABET. To address these limitations, this study introduces a novel, two‐part software platform: (1) an AI Question Generation Prompter, a web‐based tool that structures prompts based on specific course outcomes (COs) and performance indicators (PIs), and (2) a Visual Moodle XML Editor, a graphical interface that enables a “human‐in‐the‐loop” workflow for validating AI‐generated code and rendering engineering equations. A multidisciplinary exploratory case study involving mechanical, electrical, and computer engineering courses demonstrated that the platform reduced assessment creation time by an estimated 87% for one faculty participant, compared to their previous manual methods. The workflow successfully generated robust, accreditation‐aligned questions containing complex engineering notation that imported seamlessly into the Learning Management System (LMS). Furthermore, the platform operationalized a dual‐feedback loop, providing immediate formative feedback to students and data‐driven curricular insights to faculty. This work presents an exploratory, open‐source technical framework for harnessing GenAI in engineering education, ensuring both administrative efficiency and rigorous academic standards.