A Hybrid AHP–Machine Learning Framework for Integrating Generative AI as a Co-Creative Partner in Interior Design Engineering Studios

Abstract

This research focuses on the increasing but poorly specified use of Generative Artificial Intelligence in interior design engineering studios, which in the past have been focused on adopting the tool and perception-based evaluation rather than conducting engineering-level reproducible evaluation of the learning that is taking place in the co-creative processes. It evaluates a framework for assessing co-creativity using three stages (Dialogical Exploration, Critical Synthesis, and Critical Synthesis) along with an Analytic Hierarchy Process (AHP) model for weighting the co-creativity criteria, the resulting Co-Creative Performance Index (CCPI), and machine learning with explainability for predicting performance based on process behaviors. Outcome analysis from the analytic hierarchy process (AHP) indicated that the relative weights were on Creativity and Conceptual Diversity (weight w = 0.20), Critical Thinking and Justification (weight w = 0.19), and AI Literacy and Prompting Competence (weight w = 0.18), with sound expert consistency (CR = 0.06). The CCPI increased dramatically from .329 (SD = .032) prior to intervention to .599 (SD = .027), securing a mean gain of .270 (SD = .046). Predictive analysis showed that co-creative performance is a behaviorally learnable process with XGBoost achieving strong cross-validated accuracy (R^2 = 0.82, RMSE = 0.041, MAE = 0.033).