Search for a command to run...
Artificial intelligence (AI) is transforming university career services, shifting from traditional predictive algorithms to interactive generative agents. This transition raises critical questions about student agency and the digital platformisation of career guidance. In this systematic review, we examined empirical evaluations and technical architectures of AI implementations in higher education career counselling from 2015 to 2025. Following PRISMA 2020 guidelines, we identified 43 studies across Web of Science, Scopus, and ERIC. Methodologically, we appraised study quality using the Mixed Methods Appraisal Tool (MMAT) and conducted a thematic narrative synthesis. We identified three main AI implementation typologies: (1) Generative Chatbots offering scalable 24/7 support, though limited by user trust barriers and a lack of perceived empathy; (2) Predictive Analytics forecasting employability, which require transparent, explainable models; and (3) Recommender Ecosystems that match students to personalized career pathways using institutional data. Interpreting these empirical findings through Career Construction Theory and Platformisation, we highlight a tension between technological efficiency and student agency. While AI expands information access, over-reliance on opaque algorithms introduces the risk of “automation bias,” where students passively accept recommendations rather than actively constructing their vocational identities. We conclude that AI should augment, rather than replace, human counsellors within a hybrid support model. Future research should prioritize longitudinal evaluations of AI interventions, algorithmic audits of recommender platforms, and the development of Explainable AI (XAI) to foster critical AI literacy.