Two ideas came together for me this morning: a discussion of the mass graves of London, and the recent victory of AlphaGo over Lee Sedol.

No, really.

I saw this post on Tumblr: London And Its Dead. It included passages like this:

But to put that another way, the ground was so solidly packed with the interlocked skeletons of 17th-century victims of the Great Plague that the Tube’s 19th-century excavation teams couldn’t even hack their way through them all. The Tube thus had to swerve to the side along a subterranean detour in order to avoid this huge congested knot of skulls, ribs, legs, and arms tangled in the soil—an artificial geology made of people, caught in the throat of greater London.

Then there’s Wired’s discussion of the AlphaGo vs. Lee Sedol game of Go.

Like little else, this path to victory highlights the power and the mystery of the machine learning technologies that underpin Google’s creation—technologies that are already reinventing so many online services inside companies like Google and Facebook, and are poised to remake everything from scientific research to robotics. With these technologies, AlphaGo could learn the game by examining thousands of human Go moves, and then it could master the game by playing itself over and over and over again. The result is a system of unprecedented beauty.

I want you to imagine that this trend continues, and Google’s DeepMind technology is given control of another system with specific rules: the management of city infrastructure. Variables like budget, the conditions of roads and bridges, population levels, geographical constraints, and so forth.

I wrote about this in Song of Eden, postulating an automated "living city" called Hubris. In that case, the city had robots to realize its vision. But here we’ll scale back that ambition, and go with something similar. We’ll let a CityMind - a massive AI enabled by this sort of technology - essentially act as a player in a game of “SimCity”. It will make decisions about resource allocation. It will guide new construction, with input from the populace. It won’t be some sort of multi-tentacled octopus that directly runs everything - that would represent a significant strategic weakness. But its high level decisions would be fed into other systems and other processes.

Now we return to the dead of London. During the AlphaGo matches, commentators said that the AI seemed to have a “personality” or “style of play” at times. How would the London CityMind react to the presence of things like mass graves? How would its calculations include this city’s history, which spans millennia? Could the spirits of the departed metaphorically seep upwards into its calculations? And if that happened, what would it look like? Would its decisions start to seem as quirky and mysterious as the city itself?