Design Intent as a Soft Constraint: A Human-Centered Manufacturability Review Framework for AI-Generated Design Concepts

Generative AI is increasingly used in early-stage industrial design. It can quickly produce design concepts with strong visual expressiveness. However, these concepts often lack explicit dimensional parameters, material specifications, manufacturing routes, assembly definitions, and structural details. This creates considerable uncertainty for early manufacturability review. Some visual features in concept images may introduce manufacturing, assembly, or maintenance risks. At the same time, these features may also carry design intent related to appearance, functional recognition, interaction cues, or user experience. To address this issue, this study proposes Intent-Sensitive Manufacturability Review (ISMR). ISMR is a human-centered review framework that treats design intent as a soft constraint in early manufacturability review. It organizes visual cues, manufacturability risks, design-intent interpretation, potential intent loss, and required evidence for further verification in a structured manner. The proposed method is compared with an ordinary large language model (LLM)-based review and a traditional Design for Manufacturing (DFM) checklist-based review. The evaluation results show that ISMR performs more prominently in feature specificity and intent sensitivity. It can connect review issues more clearly with visible features in concept images. It also maintains attention to potential design intent when engineering risks are raised. Therefore, ISMR can serve as a useful complement to open-ended LLM review and checklist-based manufacturability review. It provides structured and intent-sensitive support for early design--engineering communication around AI-generated design concepts.