As generative AI reshapes architectural practice, a critical paradox emerges: while new computational tools offer unprecedented creative possibilities, the physical resources essential to building grow increasingly scarce. This paper investigates how architecture can productively engage this tension through novel approaches to material recycling and digital transformation. Rather than treating physical and virtual domains separately, this research proposes an integrative framework where design emerges through continuous feedback between material manipulation and computational processing. The exhibition Train Yourself demonstrates this methodology, presenting a series of experimental artifacts that evolve through iterative exchanges between physical material, AI-driven formal exploration, and imaging techniques. These hybrid objects gain definition through recursive operations that blur boundaries between physical construction and digital representation. Beyond resource conservation, this approach reveals how architectural practice might cultivate new forms of creative intelligence capable of navigating between material constraints and computational possibilities. The findings of this research suggest a mode of practice where designers orchestrate dynamic interactions between physical matter, algorithmic processes, and human intention, pointing toward a more nuanced understanding of ecological sensitivity in contemporary design practice.