Artificial intelligence and urban design

McGill University urban design studio study of urban growth in “How the other half builds”

Cities and urban settlements around the world are continuously expanding, often in ways that are to the detriment of the inhabitants of these growing cities. There are not nearly enough urban designer’s or planners to control this growth, nor are there the financial resources to pay for universal urban planning.

Designing the master plan for the Beijing Olympics’ Triathlon site, 2006

It is twenty years to the day that IBM's Big Blue computer defeated Gary Kasparov, and thus twenty years since artificial intelligence proofed that nonlinear thinking is not the preserve of the sentinel beings. What is important to realize is that it was not brute force computer calculation that allowed Big Blue to beat Gary. 

Gary Kasparov vs Big Blue

In chess, brute force cannot compete with strategic thinking because of the sheer number of variations in a chess game. There are 400 different positions after each player makes one move. There are 72,084 positions after two moves have been made. There are more than nine million positions after three moves apiece. There are close to 290 billion different possible positions after four moves apiece. By move 40, you have surpassed the number of electrons in the universe. Chesmayne calculated that there are more game-trees of chess than the number of galaxies (100+ billion), and more openings, defenses, gambits, etc. than the number of quarks in our universe!

 The computer had to learn to think like a human, with pattern recognition and by making assumptions... this was the real breakthrough with Big Blue. This is what we celebrate today.

This brings us back to the design of cities. Benjamin Franklin, America's first urbanist was a chess player. So was Napoleon, who’s cousin reshaped Paris with Haussmann's help, along urban design principals still followed today. The artist and designer, Cubist and Dadaist Marcel Duchamp was a chess player, and so was Fidel Castro, Yasser Arafat, Che Guevara, Vladimir Lenin, Tito, China’s President Chiang, Spain’s President Azana, King Abdullah the Emir of Jordan. Many nation builders, architects, urban designers and planners were chess players in their youth.

Tianjin China: Little Italy design workshop 2005

The similarities between playing chess and designing cities are striking. It is very hard to calculate the perfect urban intervention. Every decision on tracing paper has advantages and disadvantages and every urban designer has a different list of priorities and strategies. And just like every chess game played by grandmasters represents a unique set of thoughts, so is the set of thoughts unique for every new urban context under redesign. 

This leaves us with the question: where is urban design's Big Blue in all of this? Why have we not made strides in artificial intelligence in urbanism? Sure, we have GIS, GPS, satellite data, drafting software, planning databases, BIM and virtual reality, but none of these design tools represents an attempt by a computer to intervene in the configuration of human settlement. It is all analysis, and no synthesis. Not even traffic modelling qualifies as strategic thinking.

Research into this field in the 80s and 90s with a knowledge-based approach, neural networks and fuzzy logic that automates the decision-making process in planning and urbanism did not yield the same results as IBM’s Big Blue work did.

The issues determining the urban designer's decision making process are complex and many, just as is the grandmaster's considerations. Just as in chess though, an AI computer could conceivably synthesize these issues into development proposals.

Beijing China: TEDA Urban Design Workshops 2006

I am certainly not suggesting that computers replace urban designers. But in a world where the poor have yet to get full access to professional services, an AI guided urban development system for rural villages and shanty towns may improve urbanization patterns. Villages can steer clear of flood plains, shop owners don’t have to cannibalize each other’s business, commuting distances can be reduced, smart infrastructure clusters in shanty towns can bring services to millions. Shade patterns in main pedestrian spines can be optimized for warm climates, and public transit stops can be calculated based on statistical walking distances. 

Saadiyat Island Master Planning Workshop 2007

Super Computers can link data from satellites and municipalities and generate algorithms to solve problems by analyzing the interventions of human urban designers in similar contexts. All it needs is a giant database of historic human interventions in cities. It can then mine all these master plans, and urban regeneration plans for patterns. This is essentially how Big Blue beat Gary Kasparov: by analyzing the human strategic behavior in certain chess positions and by then evaluating the outcome of these strategies against positional strength algorithms. It is also how Google taught the Google earth algorithms to auto-generate three dimensional models of existing cities.

Abu Dhabi: Master planning Sir Bani yas island 2007

I have had the pleasure of practicing the dark craft of urban design from Inner Mongolia and China to Australia, from England & Romania to the UAE, from Egypt to South Africa and as far afield as Canada and Georgia, USA. As much as I cherish the memories of flying in to distant airports with my fellow urban designers, traffic engineers, environmentalists and economists to tackle developers’, mayors’ and governors' requests for growth plans and master plans, I know that we can do better. We can do better than tracing paper and Google earth printouts, Pantones, calculators and scale rulers. We are not the only ones capable of creating order out of chaos. I am reminded of this every time the computer beats me at chess!