Artificial intelligence (AI) is reshaping how work is organised and performed across the global economy. This paper reviews the evidence on three core effects: labour substitution, where AI replaces tasks previously done by people; labour augmentation, where AI raises the output and quality of human workers; and task reconfiguration, where jobs are redesigned around the new division of work between humans and machines. Drawing on peer-reviewed studies, working papers and institutional reports published between 2013 and early 2026, the review finds that outright job destruction is less common than popular accounts suggest. Instead, AI tends to substitute for specific tasks within jobs, raises productivity most sharply among less experienced workers and triggers significant redesign of roles rather than wholesale elimination. Key themes include skill compression, the emergence of hybrid human-AI roles and the uneven distribution of both risks and gains across sectors, skill levels and regions. The paper identifies gaps in the literature and sets out directions for future research, with relevance to educational institutions and policymakers in the Global South.