Abstract
The growing integration of generative artificial intelligence into non technical disciplines has highlighted a gap in how higher education institutions cultivate computational literacy among students outside computer science (CS) programmes. Current educational approaches generally fall into two extremes, they either impose syntax-heavy programming tracks that alienate non-CS cohorts or focus strictly on theoretical overviews with little technical depth. This paper proposes a multi-tier modular architecture that re-encodes the ACM/IEEE Computer Science Curricula 2023 (CS2023) standards into four functionally independent tiers: Data Foundations, Algorithmic Logic, Machine Learning and Generative Models, and Algorithmic Governance and Security. Each operationalised through no code practical instruments such as retrieval augmented generation (RAG), prompt engineering, and prompt-injection red-teaming. Using a Design Science Research (DSR) approach, the framework incorporates UNESCO's AI Competency Framework and the European DigComp 2.2 framework as a core structural layer rather than an isolated ethics topic. A Standards Traceability Matrix demonstrates one-to-one alignment between each tier, its practical tasks, and the constituent competencies of the three reference frameworks. The paper addresses four key research questions regarding architectural abstraction, technical governance, non-coding practical integration, and cross disciplinary scalability . The contribution is a reusable framework from which domain tailored AI course packages can be derived without losing technical depth, thereby repositioning governance, security, and applied comprehension as co-equal pillars of AI literacy for non CS majors. Keywords: AI curriculum design; AI literacy; CS2023; multi-tier architecture; technical governance; prompt engineering; non-CS majors; retrieval-augmented generation.