From Assistance to Substitution: Generative AI and Undergraduate Research Skills

Abstract

Generative artificial intelligence has become part of the daily academic works for the university students, whether it develops research competence or simply improves the quality and speed of immediate problem solving and response. This integrative literature review examines that question through five related dimensions of undergraduate research competence: information literacy, critical thinking and problem-solving, academic writing, self-regulated learning and research independence, and research ethics. The review synthesizes recent systematic reviews, meta-analyses, empirical studies, and established educational frameworks. However, GenAI looks most useful when it functions as an assistive tool: students use it to possibility generation, feedback receiving, alternatives comparing, or revise their work while retaining responsibility for source evaluation, reasoning, and final decisions. By contrast, substitutive use where the system performs substantial portions of the cognitive or research process creates problems of cognitive offloading, weak source studying, overconfidence, wrong references, and reduced human research independence. The review therefore argues that the central educational issue is not whether the generative artificial intelligence is beneficial or harmful in itself, but under what conditions AI-assisted performance becomes genuine, transferable research competence. Three recurring moderators are especially important: instructional directions, assessment design, and the limit of responsibility respected by the student. The review concludes with a conceptual framework distinguishing assistive from substitutive usage and proposes a research agenda centered on new designs, objective skill assessment, discipline-sensitive studies, and assessments that require students to demonstrate their own reasoning.