Claude 

Claude was used to critically review the outputs from Perplexity and ChatGPT. Instead of generating new ideas only, Claude helped compare the strengths and weaknesses of the AI-assisted process and identify which suggestions should be accepted, modified, or rejected. 

Reflection

Claude was useful because it acted as a critical reviewer rather than only generating more ideas. It helped me see that the concept of “Stillness in Motion” was strong, but the design still needed clearer structural logic. The project needed to express lightness through real architectural elements such as cantilevers, thin columns, suspended screens, counterbalance, and hovering roof planes.

This critique also revealed a weakness in ChatGPT’s visual output. Although ChatGPT helped develop the concept, the generated sketch made the building feel too solid, grounded, and fortress-like. This did not match Calder’s qualities of lightness, suspension, movement, and balance.

Claude’s limitation was that it did not design the project for me. Its feedback still needed to be interpreted through my own judgement, drawings, and design testing. This stage showed that AI can support critique and decision-making, but the final architectural direction must still come from human judgement.