To improve the design process, the outputs from Perplexity, Gemini, and Midjourney were compared and evaluated. Each AI contributed different strengths but also produced limitations that required human judgment.
Provided climate and environmental data for Kajang.
Helped identify solar exposure, wind directions, and passive design strategies.
Strong in research but limited in architectural visualization and creativity.
Developed the architectural concept and design narrative.
Helped generate ideas such as interconnected learning spaces, courtyards, and climate-responsive strategies.
Strong in concept development but occasionally produced unrealistic spatial arrangements.
Produced high-quality visualizations and architectural renderings.
Helped visualize building form, materials, atmosphere, and user experience.
Strong visually but often ignored structural feasibility and environmental performance requirements.
Human intervention was required to combine the strengths of all three AI systems.
Environmental data from Perplexity guided the design decisions, Gemini developed the concept, and Midjourney visualized the final outcome. Unrealistic forms, structural issues, and impractical ideas were refined and adjusted through architectural judgment.
The best results were achieved through a combination of AI tools and human decision-making.
AI accelerated research, concept development, and visualization, while human judgment ensured the final Contemporary Educational Building remained functional, buildable, environmentally responsive, and supportive of collaborative learning.
During the design process, differences emerged between the outputs of the AI systems.
Focused on environmental performance and realistic building requirements. It recommended strategies such as solar protection, natural ventilation, shaded glazing, and climate-responsive design solutions suitable for Kajang's tropical climate.
Focused primarily on visual impact and architectural atmosphere. It generated highly expressive building forms, large transparent façades, and dramatic architectural images that were visually appealing but sometimes difficult to construct.
As a result, contradictions appeared between environmental performance and visual aesthetics.
Perplexity (Research AI)
Midjourney (Image AI)
Climate-responsive solutions
Dramatic architectural forms
Practical shading strategies
Large exposed glazed surfaces
Realistic environmental performance
Highly idealized visualizations
Buildable design recommendations
Occasionally unrealistic structures
The final design retained the large curtain-wall glazing suggested by the visualizations but incorporated deep roof overhangs, integrated shading devices, landscaped courtyards, and naturally ventilated circulation areas.
These interventions reduced solar heat gain while preserving daylight quality and visual openness.
The building form was refined into a practical and buildable solution. Building proportions were adjusted, circulation spaces were reorganized, and façade systems were simplified while maintaining the contemporary architectural identity generated during the AI design process.
The final Contemporary Educational Building combines:
The environmental intelligence of Research AI.
The conceptual ideas of Text AI.
The visual quality of Image AI.
Human judgment reconciled the contradictions between performance, aesthetics, and constructability, resulting in a climate-responsive, functional, and realistic architectural proposal.
The final design demonstrates how Artificial Intelligence can support concept exploration and design development while human creativity, critical thinking, and professional judgment remain essential in creating meaningful architecture.
"Interconnected Form – Collaborative Learning Environment"
A climate-responsive educational building that integrates environmental performance, collaborative learning spaces, and contemporary architectural expression to create an engaging, sustainable, and human-centered environment.