A Framework for LLM-powered Design Assistants

TL;DR

The paper addresses integrating LLMs into design workflows as assistants rather than autonomous designers. It proposes a process-agnostic framework with three modalities—Idea Exploration, Dialogue with Designers, and Design Evaluation—to augment human designers. Key contributions include delineating checkpoints for AI support, detailing idea generation including cultural and historical sensitivity, dialogue facilitation for ambiguity and ethics, and evaluation across formats and comparative analyses. The work has practical significance for scalable, culturally aware, and data-driven design practices across domains like graphics, architecture, and product development.

Abstract

Design assistants are frameworks, tools or applications intended to facilitate both the creative and technical facets of design processes. Large language models (LLMs) are AI systems engineered to analyze and produce text resembling human language, leveraging extensive datasets. This study introduces a framework wherein LLMs are employed as Design Assistants, focusing on three key modalities within the Design Process: Idea Exploration, Dialogue with Designers, and Design Evaluation. Importantly, our framework is not confined to a singular design process but is adaptable across various processes.

Figures (1)

• Figure 1: The framework with three different modalities. Figure adapted from (takeda1990modeling)