Are You Optimizing For the Wrong Thing?
A conversation with Pablo Zamorano about building a technology practice centered on design values, the significance of human scale, and the implications of AI tools on creativity and the quality of architectural outcomes.
What if the biggest thing AI is doing for the architecture profession right now isn't threatening jobs, but instead making mediocrity free?
Summary
TRXL 223 is a conversation with Pablo Zamorano Mosnaim, Head of Computational Design at Heatherwick Studio in London. Pablo grew up in Chile, trained as an architect, landed in New York during the mid-2000s boom, and used the 2008 reset wisely — he moved to London, completed the AA's Emergent Technologies and Design master's program, spent a few years at SOM, and joined Heatherwick in 2015. He started as an architect working on complex geometry and eventually led the formation of the studio's formal computational design group. He's been there ever since, running technology strategy for the 260-person studio that produces some of the most geometrically specific, materially rich architecture being built right now.
The central question running through this conversation is one the industry is mostly ignoring: as AI makes it easier to produce buildings, is it making it easier to produce the right buildings, or just faster to produce more of the wrong ones? Pablo is clear about this. We are entering what he calls the era of the "ultimate easiness to produce bad architecture". It’s a moment when generating a technically adequate, geometrically unremarkable building is about to cost almost nothing. The firms that come out ahead will be the ones with ideas worth realizing and the tools and the capabilities to do it. Everyone else will find the competition has never been cheaper.
What's Inside
- The Door Scale. Why Heatherwick's design philosophy works at a small scale while most firms design for photography instead.
- Tools Leave Fingerprints. We can identify which software was used to design a building just by looking at it. What does that reveal about who is actually making the design decisions?
- The Grasshopper Shortcoming. Why Heatherwick's parametric tools saw almost no studio-wide adoption and what that taught them about building technology for designers rather than specialists.
- The AI Tool That Actually Worked. How a language-based image generation tool became the most widely adopted internal tool in Heatherwick's history within its first month.
- Values First, Tools Never Rule Them All. Why Pablo fundamentally disagrees with the one-tool-to-rule-them-all philosophy, and why Heatherwick still maintains Rhino models more detailed than their Revit counterparts well past building handover.
- They Better Find Something Else to Do. Pablo's unambiguous take on architects watching the current AI shift from the sidelines and why this moment is categorically different from previous technology transitions in the profession.
