AI as Collaborator | Closing the Loop Between Machine Learning and Human Intuition to Inform Spatial Configurations

AI as Collaborator explores a designer-in-the-loop machine-learning workflow that moves beyond text-to-image generation toward transparent, data-informed design decisions. Using designer-authored iterations, hand sketches, and labeled park layouts, a convolutional neural network learns patterns behind program placement, aiming to understand designer skillsets and propose data-informed spatial configurations. The designer remains the primary decision-maker, using model outputs to test alternatives informed by sunlight, temperature, tree canopy, pathway connectivity, and human presence. An Arduino-driven interactive 3D model displays recommendations in real time for rapid “what-if” exploration, supporting more inclusive and responsive public-space design.