Performative AI Lab
Description:
At the Performative AI Lab, which I direct at UTA's College of Architecture, Planning and Public Affairs, my research explores artificial intelligence as a generative and analytical instrument for advancing spatial, ecological, and experiential performance in architecture. I study how AI mediates form, performance, and environmental responsiveness across scales—from material and tectonic logics to the dynamics of urban ecologies—developing computational design frameworks for decarbonization, resilience, and environmental performance.
The lab pairs a 25-foot curved, AI-enabled LED video wall with an advanced robotic fabrication system, supporting immersive modeling, AI-feedback integration, and the materialization of AI-generated artifacts.
Two projects currently drive this research:
Agentic AI
Agentic AI develops and tests AI-driven design frameworks in which generative models function as design agents — large language models and diffusion (image-generating) models trained on concept-specific ecological design strategies. Rather than serving as representational tools, these models operate as an autonomous design intelligence, proposing and evolving architectural and urban interventions in dialogue with the designer.
The framework privileges ecological objectives as its generative logic, so that performance, adaptation, and resilience become the criteria that shape form. Researchers and students train their own models on targeted architectural datasets, and the designer's role shifts toward curating, steering, and refining the possibilities these agents generate toward ecologically grounded outcomes.
Materializing AI Ecologies
Materializing AI Ecologies turns from representation to matter, examining how artificial intelligence catalyzes generative and performative design at the architectural and spatial scale. Robotic fabrication becomes the medium through which algorithmic intelligence is transposed into built artifact — the threshold where computation meets material behavior.
The project cultivates a co-creative framework in which machine learning augments design authorship and material responsiveness, blurring the line between designer and system. Its outcomes are interactive spatial prototypes that embody behavioral intelligence and ecological attunement — physical demonstrations of a design practice that senses, adapts, and responds.