Artificial intelligence is reshaping both the objects that information systems researchers design and the way design research is conducted. Existing literature offers little guidance for computational design science (CDS) when AI actively participates in problem formulation, resource construction, design search, evaluation, and knowledge abstraction. To address this gap we introduce AI4CDS (AI for Computational Design Science), a five‑phase methodological framework: 1) problem expansion; 2) resource generation; 3) design space exploration; 4) result evaluation; 5) knowledge abstraction. Researchers retain responsibility for domain grounding, admissibility verification, and scientific judgment, while AI expands the search space. Collaboration is governed by graduated trust, reversibility, auditability, and differentiated reproducibility. We instantiate AI4CDS with ChildRiskGuard, an interpretable artifact that detects short‑form videos unsuitable for children. The case documents AI‑researcher interactions, rejected alternatives, corrections, and audit trails. Audience‑dependent safety and explanation faithfulness are translated into three technical challenges: separating generic from child‑specific risk, representing distinct developmental‑risk mechanisms, and integrating concept‑level explanations into the predictive computation. ChildRiskGuard maps risk factors to concept vectors and emits corresponding explanations during inference. Empirical results show an F1 score of 0.769, substantially outperforming a direct application of a general‑purpose content‑safety model and remaining competitive with strong benchmarks. The primary contribution is AI4CDS as a responsible framework for AI‑enabled CDS; ChildRiskGuard provides process and artifact evidence that researcher‑governed, AI‑expanded design can generate and evaluate novel computational design knowledge.
Review