I was invited, last Friday, to be part of a panel discussing AI and the city. There are quite a few such panels these days. The common question seems to be: how much will AI alter cities?
Not much, I think (and, incidentally, Paul Krugman would probably think so too).
AI is about data processing, analysis and use
AI is a continuation of developments in computing, data gathering, data management, and data analysis that have been gathering momentum since the cybernetics of World War II. It combines information processing, big data and machine learning: AI will have impact in these areas.
And, as its impacts ripple out, it will affect other fields, but these impacts will diminish as the fields are further removed from information processing.
Cities are not information, big data or computers. Previous information processing revolutions – such as the printing press, television, internet etc… have indirectly impacted cities, but not in the short term, nor in any simple way.
What has impacted cities directly and fairly quickly?
It is metalled roads; sewers; tall buildings; elevators; energy sources and usage (such as oil and the internal combustion engine); water and food distribution; political processes; regulations; governance structures; developers; communal action; etc… that have fashioned and changed cities.
Cities are material entities
Cities are, of course, congregations of people (who, amongst other things, exchange information); but this does not define cities: virtual communities, communities of belief, political parties, choirs and sports teams can also be described this way.
What distinguishes ‘urban’ communities is that they are indissociable from their material setting and from physical processes that require a material substrate.
What makes cities ‘cities’ is their materiality: the bricks, steel and concrete that make their buildings; the asphalt, gravel and stones that make their roads; the trucks that deliver food, and goods, and that dispose of waste; the water processing plants; the reservoirs; the conduits that bring water and dispose of it; the trees that line streets; the city halls where people meet; the sidewalks and parks they walk on; the fibre optics that deliver Internet (and Chat GPT); the electric cables that provide power … and the social and political systems that allow these material processes to be imagined, financed and coordinated.
None of these will be much affected by AI, though some may be optimised or slowly altered as an indirect effect.
AI may optimise processes that should be fundamentally changed
AI will no doubt optimise, at the margin, some functions that urban communities rely upon, such as:
- improving waste collection (but shouldn’t we reduce waste?);
- smoothing traffic flows (but shouldn’t we reduce traffic?);
- altering governance processes (but the real challenges are that there are only 24 hours a day, that interest groups can easily take-over these processes, that power is at play…);
- informing us when the next bus is due (but that is what timetables did, before congestion and public service disinvestment made them unreliable…);
- matching renters with apartments (but we need more and more affordable housing, not an AI optimised leasing system).
AI optimization may prevent fundamental change as undesirable environmental, social and economic processes are perpetuated rather than reimagined
AI will also increase surveillance and targeted marketing – this is not uniquely urban, but will affect cities unless legislators begin to harness and regulate it. But again, this is not a revolution, but a sad continuation of neighbourhood watch, CCTVs and of certain types of policing.
AI will not :
- solve the housing crisis;
- fill potholes;
- mend crumbling bridges;
- ensure there is health-care for all;
- reduce gun violence;
- reduce drug overdoses;
- stop certain drivers from speeding, certain cyclists from running red lights, or certain pedestrians from stepping in front of bikes….
- ensure all voices are heard and respected;
At best AI may predict when these things are likely to flare up, describe these issues (relying on murky sources and algorithms), and let urban managers and politicians know that these issues need solving – but they know that already.
Indeed, it seems that AI will not address any of the pressing urban issues of the day.
AI is nothing without wires and well-tended trees
But could AI have prevented Wednesday’s ice storm in Montreal? No. At best it may have marginally improved the weather forecast (the storm was amply forecast).
The storm (which has left over 500 000 Montrealers without power for about two days) has starkly reminded us that all the wonders of AI are nothing without wires bringing electricity to our homes, fridges to keep food fresh, heating to keep warm, and battery powered radios to pick up a little news.
Oh yes – well-tended trees are also pretty important.
Richard Shearmur 