AI Sampling Singapore: Redefining High-Rise Housing Design

Immanuel Koh’s AI-driven project imagines the future of Singapore’s urban architecture.

Singapore’s reputation as an “artificial city” serves as the conceptual foundation for AI Sampling Singapore, a pioneering project by Immanuel Koh. By leveraging a vast dataset of 3D digital models of the city’s high-rise buildings, the project’s custom AI model not only replicates existing architectural typologies but also invents new, unexpected forms. These generative outputs interpolate and extrapolate among familiar slab, point, and cluster block structures, resulting in configurations that are both plausible and strikingly novel. The project’s significance has been recognized internationally, with exhibitions at the Venice Architecture Biennale and Singapore’s Arts House.

At its core, AI Sampling Singapore investigates the potential of deep generative neural networks to learn and reimagine the three-dimensional exterior and interior spaces of public housing. Utilizing a redesigned 3D generative adversarial network (3D-GAN), the AI model is trained to understand the semantic and spatial logic of Singapore’s high-density residential architecture. This approach enables the creation of new buildings that align with local urban patterns, yet push the boundaries of conventional design. The project’s innovative methodology has earned accolades, including the WAFX2024, IDA2024, and BLT2024 awards.

The realization of this digital architecture relies on a proprietary codebase and a meticulously curated dataset. One of the primary challenges addressed by the team was the lack of comprehensive 3D models and floor plans for Singapore’s housing stock. To overcome this, new workflows and tools were developed for data collection, annotation, and exploratory analysis. The technical complexity of scaling such a dataset was managed through a novel AI latent representation, allowing the model to generate high-resolution, 30-floor residential towers.

AI Sampling Singapore is not just a design exercise but a long-term research initiative conducted at a university design lab since 2021. The project’s digital outputs are grounded in real-world data, yet the AI’s creative agency enables the emergence of forms that transcend traditional regulatory frameworks. This raises important questions about the future of architectural practice and the role of AI in shaping urban environments.

As a recipient of the Iron A’ Design Award in 2025, AI Sampling Singapore exemplifies the integration of advanced technology with practical, industry-relevant solutions. The project highlights the transformative potential of AI in architecture, offering a glimpse into how cities like Singapore might evolve through the synergy of data, design, and machine intelligence.

AI Sampling Singapore stands as a testament to the possibilities unlocked when artificial intelligence is harnessed for architectural innovation. By reimagining the city’s skyline, the project invites architects, urban planners, and technologists to consider how generative design can contribute to more adaptive, locally resonant, and visionary built environments.