Artificial Intelligence (AI), predictive analytics, and urban datarization are fundamentally reshaping how planning knowledge is produced, contested, and applied. While these technologies promise efficiency and evidence-based decision-making, they also risk narrowing the epistemological foundations of planning privileging algorithmic rationality over locally embedded, participatory, and socially responsive ways of knowing. This tension is especially acute in the Global South, where rapid urbanization, widespread informality, and uneven digital infrastructures complicate the uncritical adoption of data-driven planning models predominantly developed in Western contexts. Drawing on Foucault's concept of power/knowledge alongside spatial justice frameworks, the paper situates AI-driven planning within broader struggles over whose knowledge counts, who governs urban space, and on whose terms. It argues that smart-city paradigms, as currently conceived, risk reproducing colonial and technocratic logics that sideline informal realities, local expertise, and community voice precisely the conditions most prevalent across Global South cities. Responding to the conference theme of Building Planning Up, this paper proposes a spatial justice-oriented framework for ethical, participatory AI-enabled planning. It contributes to debates on theoretical renewal, methodological innovation, and disciplinary identity arguing that critically re-examining planning's intellectual foundations is essential to ensuring the discipline remains inclusive, reflective, and fit for purpose in an age of artificial intelligence.