Personalized learning paths (PLPs) have emerged as a promising strategy to enhance student outcomes. These paths tailor educational content, pace, and assessment methods to meet individual student needs and learning styles, promoting deeper engagement and better knowledge retention. In this paper, we survey and analyze existing solutions, both AI-driven and non-AI approaches. AI-based methods leverage techniques such as machine learning, natural language processing, and recommender systems to dynamically adapt learning experiences. At the same time, non-AI solutions rely on rule-based systems, instructor-driven customizations, and predefined learning sequences. By comparing these approaches, we highlight their strengths, limitations, and applicability in the university STEM context. Our analysis aims to provide researchers, educators, and policymakers with insights into the current landscape of personalized learning path development and guide future research directions for integrating AI responsibly and effectively in higher education.