AI Research

AI for Computational Design Science: A Responsible... | AI Research

Key Takeaways

AI is fundamentally changing how researchers conduct design science, yet there has been a lack of clear guidance on how to integrate AI into the research pro...
Artificial intelligence (AI) is transforming not only what information systems researchers design, but also how design research is conducted.
Yet existing literature offers limited guidance for computational design science (CDS) when AI actively participates in problem formulation, resource construction, design search, evaluation, and knowledge abstraction.
Collaboration is governed by graduated trust, reversibility, auditability, and differentiated reproducibility.
We instantiate AI4CDS through ChildRiskGuard, an interpretable artifact for detecting short-form videos inappropriate for children, while documenting AI interactions, rejected alternatives, corrections, and audit trails.

Paper AbstractExpand

Artificial intelligence (AI) is transforming not only what information systems researchers design, but also how design research is conducted. Yet existing literature offers limited guidance for computational design science (CDS) when AI actively participates in problem formulation, resource construction, design search, evaluation, and knowledge abstraction. We develop AI for Computational Design Science (AI4CDS), a five-phase methodological framework in which AI expands problem and design search while researchers retain responsibility for domain grounding, admissibility, verification, and scientific judgment. Collaboration is governed by graduated trust, reversibility, auditability, and differentiated reproducibility. We instantiate AI4CDS through ChildRiskGuard, an interpretable artifact for detecting short-form videos inappropriate for children, while documenting AI interactions, rejected alternatives, corrections, and audit trails. The case translates audience-dependent safety and explanation faithfulness into three technical challenges and develops an artifact that separates generic from child-specific risk, represents distinct developmental-risk mechanisms, and makes concept-level explanations part of the predictive computation. ChildRiskGuard achieves an F1 score of 0.769, substantially outperforming direct application of a general-purpose content-safety model while remaining competitive with strong benchmarks. The primary contribution is AI4CDS as a responsible framework for AI-enabled CDS; ChildRiskGuard provides process and artifact evidence of how AI-expanded, researcher-governed design can generate and evaluate novel computational design knowledge.

AI is fundamentally changing how researchers conduct design science, yet there has been a lack of clear guidance on how to integrate AI into the research process responsibly. This paper introduces a new methodological framework called AI for Computational Design Science (AI4CDS). The goal is to allow AI to actively participate in complex research tasks—such as problem formulation and design search—while ensuring that human researchers remain in control of scientific judgment and verification.

A Framework for Human-AI Collaboration

The AI4CDS framework consists of five phases designed to balance AI’s computational power with human oversight. In this model, AI is used to expand the scope of problem-solving and design exploration. To ensure the process remains responsible, the collaboration is governed by four key principles: graduated trust, the ability to reverse actions, full auditability of the process, and differentiated reproducibility. By following this structure, researchers retain responsibility for domain grounding, verifying the results, and ensuring the admissibility of the design. The same ai evaluation question is explored in A Computationally Feasible Framework for Causal..., which adds a research perspective.

Putting the Framework to the Test: ChildRiskGuard

To demonstrate the effectiveness of AI4CDS, the authors developed an artifact called ChildRiskGuard. This tool is designed to identify short-form videos that are inappropriate for children. The researchers used the AI4CDS framework to guide the development process, carefully documenting every interaction with the AI, including rejected design alternatives and necessary corrections. This documentation serves as an audit trail, proving how the human-AI collaboration evolved throughout the project.

Technical Innovation and Performance

ChildRiskGuard addresses the challenge of distinguishing between generic content safety and specific risks to children. The artifact separates these two types of risks and incorporates concept-level explanations directly into its predictive computation, making the system more interpretable. In testing, ChildRiskGuard achieved an F1 score of 0.769. This performance significantly outperformed general-purpose content-safety models and remained competitive with other strong benchmarks in the field. The same ai evaluation question is explored in WikiSkill, which adds a research perspective.

Advancing Design Science

The primary contribution of this work is the AI4CDS framework itself, which provides a roadmap for how researchers can use AI to generate and evaluate new computational design knowledge. By documenting the entire process—from the initial AI interactions to the final artifact—the authors provide evidence that AI-expanded, researcher-governed design is a viable and effective path for future scientific research. The same ai evaluation question is explored in Does Your Agent's Memory Survive a..., which adds a research perspective. as detailed in the full paper on Arxiv

Comments (0)

No comments yet

Be the first to share your thoughts!