Gemini

Gemini was used as a third AI judge after Perplexity, ChatGPT, and Claude. This step was important because the AI tools did not all play the same role. Perplexity provided site and climate research, ChatGPT developed the poetic concept of “Stillness in Motion,” and Claude challenged the project by questioning whether the concept had enough structural logic. Gemini was then used to compare these positions and help decide which ideas should be accepted, modified, or rejected before moving into the revised design proposal.

This helped make the AI workflow more critical. Instead of depending on one AI answer, I used Gemini to check where the tools agreed, where they disagreed, and which direction was most convincing for the project.

Conclusion

Gemini’s review showed that the project should not reject ChatGPT or Claude completely. ChatGPT was useful for developing the conceptual language of balance, suspension, light, shadow, and movement. However, Claude’s critique was stronger when it came to architectural reality because it explained that lightness must be shown through structure, not only described in words.

From Gemini’s judgement, the final direction became clearer. The revised proposal should keep ChatGPT’s concept of “Stillness in Motion,” follow Claude’s demand for a clear structural gesture, and maintain Perplexity’s climate strategies for shading, ventilation, and sea orientation. This means the design should move toward a realistic “Coastal Stabile”: a stable base, a lighter steel structure above, a tapered roof gesture, an off-center courtyard, and fixed blade-like louvers that create shade, airflow, and moving shadows.

This stage showed that AI comparison is useful because it helps reveal contradictions between concept, climate, structure, and realism. The final decision still depends on human judgement, because I had to choose which AI ideas fit the site, the client, and the architectural intention.