The rapid incorporation of algorithmic systems into governance has raised expectations that the Indian state can make public decision-making more objective and efficient. These developments coincide with ongoing debates on affirmative action, caste-based inequalities and the constitutional mandate to ensure substantive equality. Yet, technological systems, however sophisticated, remain limited by the nature of the data on which they operate and the assumptions embedded in their design. This paper argues that the goals of affirmative action under the Constitution of India are best served through a balanced, human-centred socio-technical model that combines the strengths of AI with the indispensable capacities of human judgment, such as empathy, contextual understanding, ethical reasoning and interpretive sensitivity. Drawing from constitutional theory, social research on caste and structural inequality and jurisprudence, particularly the Supreme Court’s 2024 decision in State of Punjab v. Davinder Singh, the paper highlights how AI can enhance administrative efficiency while humans ensure that decisions remain grounded in constitutional morality and social realities. Rather than positioning technology as a substitute for human intellect, the analysis proposes a collaborative framework in which AI operates as a supportive tool that augments, rather than replaces, human decision-making. Such an approach aligns with the constitutional vision of justice in India, where technology contributes meaningfully to equity outcomes while final judgement remains attentive, humane and morally anchored...