Prof. Dr. Ualsher Tukeyev,
Faculty of Information Technology and Artificial Intelligence, Al-Farabi Kazakh National University, Kazakhstan
COMPUTATIONAL THINKING AS CREATIVE MODELING IN AI-ASSISTED INTELLIGENT PROGRAMMING
AI-assisted programming tools and large language models have fundamentally changed software development. Today, AI can generate full codebases, suggest architectures, create tests, and even assemble deployable systems from high-level prompts. While this dramatically accelerates implementation, it does not eliminate the need for structured problem formulation and model-level reasoning.
This plenary lecture proposes a modeling-centered interpretation of computational thinking in the era of AI-assisted development. Rather than equating computational thinking with coding, we argue that its constructive phase lies in creative modeling—the explicit design of executable computational models grounded in domain constraints, representation choices, and validation criteria.
In practice, creative modeling involves:
(1) defining the problem space and non-negotiable constraints,
(2) selecting or designing appropriate computational representations that explicitly encode domain structure and constraints,
(3) constructing hybrid executable models, and
(4) specifying validation procedures independent of implementation details.
AI-assisted programming then operates as an implementation accelerator within this structured workflow.
The approach is illustrated through a hybrid symbolic–neural model of Kazakh morphological structure, where constraint-aware modeling ensures structural correctness while neural components enableplenary generalization. Experimental results demonstrate that explicitly modeled constraints improve robustness and reduce structurally invalid outputs compared to unconstrained neural solutions.
The lecture concludes by outlining when creative modeling becomes essential: in constraint-sensitive systems, hybrid architectures, explainable AI applications, and validation-critical domains. In such contexts, intelligent programming requires not only code generation but explicit model construction and verification.
This perspective reframes the role of the human engineer: from code producer to model architect responsible for structural integrity and correctness in AI-driven systems.
