Abstract. The digitization of higher education has intensified the need for scalable, pedagogically grounded solutions to deliver hands-on, lab-based learning in online environments. This paper presents a five-layer framework for designing AI-powered virtual labs with authoring tools that integrate constructivism, Bloom’s Taxonomy, the Watch-Try-Do instructional model, AI-driven tutoring via LLM/RAG, xAPI-enabled learning analytics, and scalability and data ethics into a coherent implementation roadmap. The framework is instantiated through an Excel lesson on formulas that operationalizes all layers using a Watch-Try-Do sequence, an embedded GPT-4o-mini chatbot, and xAPI tracking to a Learning Record Store. A formative mixed-methods evaluation with ten staff at the University College of Applied Sciences in Gaza examined implementation outcomes (Acceptability, Appropriateness, Feasibility, Concerns) and adoption conditions. Reliability was satisfactory to excellent across constructs, and descriptive statistics with one-sample Wilcoxon tests showed Appropriateness and Feasibility to be above a neutral benchmark, while Acceptability was positive but sensitive to conservative inference and Concerns were heterogeneous rather than uniformly high. Thematic analysis of open-ended responses surfaced key benefits, barriers, enablers, and needs for institutional adoption. Triangulated mixed-method findings position the framework as pilot-ready under phased implementation and identify critical enabling conditions required to sustainably scale AI-powered Vlabs in higher education. The study contributes to a transferable model, an instantiated use case, and initial empirical evidence to guide the design and rollout of AI-powered virtual labs in higher education.
Full citation. M. J. Almasri, A. S. Imran, A. Mavroudi and A. Soylu, "AI-Powered Framework for developing Virtual labs in Higher Education," 2026 IEEE Global Engineering Education Conference (EDUCON), Cairo, Egypt, 2026, pp. 1-10, doi: 10.1109/EDUCON67543.2026.11574439.