<![CDATA[This study investigates the effectiveness of a project-based learning framework in scaling the design cognition of architecture students through structured integration of generative AI tools, examining whether a tiered pedagogical approach helps students transition from passive tool users to active directors of machine intelligence. The research employed a project-based strategy merging the Stanford Design Thinking model with the Educational Design Ladder. Seventy-three architecture students from Universitas Islam Indonesia participated in the Computational Design Thinking course during 2025/2026. Students progressed through manual nature observation, prompt engineering with language models, generative image creation, 3D model conversion, parametric refinement, and physical fabrication. Data were collected through surveys, grading rubrics measuring higher order thinking skills, student reflections, and comparative analysis with non-AI cohorts. Results showed measurable improvement, with average grades rising from 75.3 to 77.9 compared to previous non-AI cohorts. Significant gains occurred in translating ideas into computational logic (+6.8 points) and design report quality (+6.9 points). Students reached Synthesis and Evaluation levels of the Educational Design Ladder (79.2). Self-reported cognitive habits improved from 3.60 to 3.71, while prompt writing proficiency increased from 3.48 to 3.63. Students generated over ten design alternatives per concept, demonstrating expanded creative output. Physical 3D print quality showed slight decline, and students reported cognitive load during transitions to node-based interfaces and hardware installation barriers. Single institution study over one semester limits generalizability. Technical barriers affected some participants. Focus on early-stage design cognition may not capture long term skill retention. Future implementations should introduce computational tools earlier in the curriculum, provide robust hardware infrastructure and technical support, and conduct longitudinal studies on professional application. This research provides a replicable pedagogical roadmap for integrating AI while maintaining human creative authority. Institutions should embed AI throughout the curriculum, establish ethical guidelines, and provide structured technical scaffolding to help students achieve higher order thinking skills in human AI collaboration.]]>
