Bridging Undergraduate and Graduate Research Training in the Era of Generative Artificial Intelligence: A Progressive Human-AI Collaborative Framework for Engineering Education

Abstract

Generative artificial intelligence (GenAI) is lowering the procedural barriers to literature synthesis, coding, data analysis, simulation support, and scholarly communication, thereby creating new opportunities for introducing authentic research earlier in engineering education. Yet easier access to research-like outputs does not necessarily produce research competence, and undergraduate students may enter graduate study with uneven abilities in problem formulation, methodological justification, evidence verification, uncertainty reasoning, and academic responsibility. This design-oriented study develops a progressive framework for AI-enhanced undergraduate research training that explicitly targets graduate-research readiness. Drawing on outcome-based education, constructive alignment, undergraduate research scholarship, and the engineering-education principles of authentic problems, project-based learning, numerical modeling, simulation, and evidence-based assessment, the framework organizes training into four stages: research and AI literacy, guided inquiry, semi-independent research, and a graduate-entry bridge. It further proposes a faculty mentor–graduate near-peer–undergraduate–GenAI collaboration mechanism in which support is distributed but epistemic ownership remains with the undergraduate researcher. For implementation, the study specifies an AI-supported research task chain for mechanical and vehicle engineering and develops a closed-loop readiness assessment based on traceable process evidence rather than the polish of a single AI-assisted product. The proposed approach reframes undergraduate research from isolated participation toward calibrated research independence and provides a practical architecture for improving continuity between undergraduate research experiences and graduate-level research training.