Algorithms and artificial intelligence (AI) are often portrayed as powerful tools intended to enhance performance and optimize tasks. Yet critical scholarship stresses that these systems operate as socio-technical actors: they are conceived, trained, and deployed within particular institutional and social arrangements, and their outputs may generate distortions in the very domains they are meant to improve. Spatial discrimination—an enduring phenomenon—emerges when resources, opportunities, or services are allocated in ways that sustain or intensify geographic inequalities, sometimes through apparently neutral criteria that nonetheless disadvantage specific places or populations. This study offers a systematic literature review that examines how AI may interact with territorial dynamics and how spatial inequalities can be reproduced through algorithmic decision-making. It identifies two key dimensions through which territorial discrimination can take shape, helping to frame where discriminatory effects may originate and how they may be experienced across space. This paper argues that a critical approach to algorithm implementation is essential to ensure ethical, transparent, and rigorous assessment. Such an approach requires careful scrutiny of underlying assumptions, data sources, proxies for location, and decision rules, as well as attention to governance and accountability in real-world deployments. Overall, this study underscores that AI is not merely a technical instrument but also a force that can reshape spatial governance and influence whether geographical inequality is reinforced or reduced over time.