<![CDATA[The integration of Artificial Intelligence (AI) into legislative processes offers opportunities to improve information management and drafting efficiency, but raises questions about constitutional legitimacy and economic feasibility. This study aims to examines the legal boundaries and efficiency conditions governing AI use in Indonesian law formation and develops an AI-Assisted Legislative Process model. Using normative juridical methods with statutory and conceptual approaches, it analyzes constitutional provisions, legislation, and the institutional context of the AKSI platform and Data Sharing Portal developed by the Expertise Body of the House of Representatives of the Republic of Indonesia. The findings indicate that AI cannot replace constitutionally authorized law-making institutions under Articles 20 and 22D of the 1945 Constitution and must remain a human-controlled decision-support tool. From an Economic Analysis of Law perspective, AI adoption is efficient only when information-cost savings exceed transaction, verification, error, and social costs, consistent with the Kaldor–Hicks efficiency criterion. Accordingly, the study proposes a nine-stage AI-Assisted Legislative Process model incorporating two control mechanisms: an EAL Cost-Benefit Gate and a Constitutional Legitimacy Check. The study proposes an integrated framework for evaluating AI-assisted legislation and recommends human-centered safeguards and binding regulations for AI use in legislative processes.]]>
