Generative AI in Professional Spatial Design: The D-E-T Conceptual Framework for Human-AI Co-evolution
Year: 2026
Type: preprint
Abstract: Generative artificial intelligence can already support designers in organizing project information, exploring design directions, and generating outputs, yet generative capacity alone does not amount to professional design capability. Current practice still faces three structural problems: the relationship between AI generation and human judgment remains insufficiently structured; general-purpose models do not consistently provide the expertise required in professional spatial design; and complex design intent is difficult to translate reliably into tasks that AI can execute. Following a model-building approach to conceptual research, this paper proposes the D–E–T framework, comprising Decision Governance, Expertise Extension, and Intent-to-Task Translation, with project context and stage memory as their shared foundation. The framework uses a project loop to support the joint evolution of problems, solutions, and evaluation criteria, and a capability loop to accumulate reusable decision criteria, professional expertise, and translation methods across projects. Its contribution is to place decision authority, the source of professional expertise, and task translation within a unified, testable framework for human–AI design collaboration.
Author: Jason Leo, Zhao
Language: English
Subfield: Architecture
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7204862331
Cites:
Cited by:
Related to: