Your computer could learn to design a city?
Self-driving cars were once the realm of science fiction but now have become a normal everyday expectation of how transportation and engineering are evolving. But what about architecture and city design? Could computers learn to do such a complicated set of tasks? For anyone who has played city-building simulators like SimCity from Electronic Arts (NASDAQ: EA) or Cities: Skylines, computers have had a fair bit of ‘experience’ at city-building in the virtual world – if only at a toddler-grade level.
But, perhaps that is about to change. Autodesk (NASDAQ: ADSK), mainly known for its architecture design software, is using its AI-based tool to offer possibilities for designing a city block. Simply put, plug in the objectives that you want to prioritize, such as traffic flow, environmental sustainability, property value, population density, and the computer shoots out various options and suggestions on where to put that fountain, or how many trees your park really needs, or how to organize your roads, so you don’t cause bumper-to-bumper traffic at 4 pm. Well, sort of.
Right now, these AI tools are just in their building-block learning phase – really, the computer is relying on humans to take or reject suggestions that it spits out. This is not unlike self-driving cars, which need astronomical amounts of learning examples and data to program the software. But it is certainly a testament to the power of computers and AI to learn the tricks of the trade.
One might ask, one day, will we even need human architects anymore for designing residential and commercial buildings? Well, based on the current level of ‘aesthetic’ intelligence that’s demonstrated by computers that try to generate graphic art, let’s just say I wouldn’t be putting in any bids on those any time soon.
It’s fair to say that designing anything is an entirely different category of activity, than driving. For self-driving cars, there are objective rules that can be followed repeatedly in simulated environments and the results of that training can be applied on any real road. Structures and cities are unique in time and space, and each design, each creation, requires a human eye to see it through to completion. Getting it wrong one or many times leads to a dreary, dull, or ugly environment no one wants to look at or live in.
Whatever that X-factor is that humans have that computers don’t quite have yet, one thing is for sure: these new AI tools are helping architects, designers, and everyday folk to learn what future cities could look like. And, like good science fiction or simulator games, that could help us to reimagine a better future.