Published Paper Details:

Large Language Model (LLM)-based chatbots such as ChatGPT, Gemini, and Copilot are rapidly being integrated into higher education and professional learning environments for tutoring, assessment support, and content generation. While these tools demonstrate strong potential to enhance personalized learning, scaffolding, and accessibility, concerns persist regarding overreliance, reduced cognitive engagement, and erosion of critical thinking skills. This study critically examines whether educational chatbots function primarily as cognitive supports or cognitive substitutes. Using a systematic review of global empirical studies, policy reports, and experimental findings, the paper analyzes learning outcomes, student behavior, and pedagogical integration strategies. Secondary data are synthesized through comparative tables highlighting learning gains, engagement metrics, and cognitive risks. Findings suggest that chatbots enhance conceptual understanding and feedback efficiency when used as metacognitive tools but may suppress higher-order thinking when used for answer generation and task completion. The study concludes that learning outcomes depend less on the technology itself and more on pedagogical design, assessment strategies, and institutional policy frameworks. The paper proposes instructional design principles and governance guidelines for responsible AI integration in higher education.