Making the Implicit Explicit: A Human-In-The-Loop AI Pipeline for Excavating and Making Use of Latent Design Knowledge
Abstract
Design science research (DSR) increasingly depends on engaging with large bodies of artifact-oriented research beyond the ‘traditional’ DSR literature. Yet, extracting and formalizing design-relevant insights from non-DSR publications remains highly manual because prescriptive intent is typically embedded in technical descriptions, evaluation results, and trade-off discussions rather than expressed as reusable design knowledge. By following an exploring-by-building design logic, we present AIDE. AIDE is a human-in-the-loop, AI-assisted pipeline that excavates latent design knowledge from non-DSR corpora and transforms it into formalized representations. We instantiate AIDE and demonstrate its feasibility in an interdisciplinary corpus on AI-enabled circular economy, reconstructing one structured design principle per study under expert validation. The paper contributes (1) a method for design knowledge excavation, (2) an instantiation that
operationalizes transparent human–AI collaboration, and (3) a broadened perspective on DSR as an integrating meta-discipline that can systematically draw on latent knowledge across disciplinary boundaries at scale.
Keywords
Design Grounding; Design Knowledge; Latent Design Knowledge; Human-AI Collaboration; Computational Design; Science Research