AI-Driven Mentorship in Education: A Systematic Review of Educational Chatbots, Stress Detection Systems, and Career Guidance Platforms

Abstract

Artificial Intelligence (AI) has emerged as a promising enabler of scalable and personalized mentorship in education, addressing limitations of traditional human-centric mentoring models such as restricted availability, inconsistent quality, and high student-to-mentor ratios. This systematic review synthesizes 134 peer-reviewed studies published between 2018 and 2025, selected using PRISMA 2020 guidelines from IEEE Xplore, ACM Digital Library, SpringerLink, Scopus, and PubMed. The review critically examines three dominant application domains: (i) educational chatbots and intelligent tutoring systems, (ii) AI-based stress and well-being detection mechanisms, and (iii) AI-powered career guidance platforms. Findings indicate that educational chatbots effectively support cognitive learning outcomes but rarely incorporate emotional awareness or long-term mentoring functions. Stress-detection systems demonstrate high predictive accuracy in controlled or laboratory settings but lack deployment within authentic educational environments and rarely trigger mentoring interventions. Career guidance platforms offer scalable and structured pathway recommendations but operate independently of students’ emotional states and academic performance. Across all domains, research remains fragmented, pilot-driven, and weakly integrated, with limited longitudinal validation, minimal focus on transferable skill development, and insufficient ethical governance. This review consolidates existing evidence, identifies systemic research gaps, and outlines future research directions necessary for advancing holistic, ethical, and student-centered AI-driven mentorship in education.