Human Judgment + Output Evaluation
Evaluating AI Outputs Through Architectural Thinking
Embedded Files
Human judgment was necessary in every stage of the workflow. AI supported the process by generating research, concept text, and visual ideas, but the final design decisions remained with the designer.
Some AI outputs were useful, while others needed to be edited, refined, or rejected. Gemini provided useful research about tropical design strategies, but some information was general. ChatGPT helped organize the concept narrative, but the text needed to be adjusted to match the project. MidJourney generated strong visual atmosphere, but some images were not fully realistic or buildable. Revit and diagrams helped test the ideas through architectural logic.
This stage shows that AI can support architectural thinking, but it cannot replace the architect’s responsibility. The designer must evaluate the output, understand the site, check the design logic, and decide what is suitable for the project.
AI can generate options, but the architect must decide what is meaningful, realistic, and suitable.
Figure 13A: Human judgment was used to evaluate AI outputs and refine the final architectural workflow.
Figure 13B: Sectional perspective and exploded view showing how the concept was developed into architectural logic and details.
Key Principle
Key Principle
AI assists the workflow. Architects remain responsible for the decisions.
Page updated
Google Sites
Report abuse