OK, so we can expect a smart city to become more and more “imbibed” with IoTs, hundreds of millions of them connected through a nervous infrastructure supported by a ubiquitous communications fabric (5G) with plenty of software that makes sense of data and orchestrate the response of the city, seen as a living organism, to a changing environment. This city, thanks to this “nervous soft infrastructure” will be getting smarter and smarter, it will be capable of learning from past experiences and adapt to the variety of situations arising.
Will this city be intelligent? I personally doubt it. The amount of IoT and their mutual interconnections via the soft infrastructure seems well below the level of complexity that would be required for an autonomous emerging intelligence. You can have intelligent systems plugged in providing specific intelligent responses, but the city is unlikely to become intelligent, as a whole.
If you look at animals the number of IoT we can expect in a 100,000 people city would compare to the number of neurones in a mouse brain BUT the number of interconnections in a mouse brain is in the order of 1 trillion and that is way, way, beyond what we have in a city of that size (the connectivity of a city with one hundred thousands inhabitants may compare to the interconnections present in a honeybee brain, in the order of billions). We may feel that a mouse is intelligent (to a certain extent) but I would say we all agree that a bee is “smart”, pretty smart, but not intelligent.
Of course, if it can make sense, up to a certain point, the comparison of a neurone with an IoT (although many IoT do not have the processing and the memory capability of a neurone) it is much more tricky to compare the synapses (interconnections in a brain) with the soft interconnectivity structure of a city, based on software and data centres (these, in fact, replicate part of the functions of neurones and neuronal circuits, like processing and memory/experience).
Here, however, I am just trying to reflect on general aspects of emerging intelligence, so I guess we can use these loose comparisons.
Is there a chance for a city to reach the level complexity that we see in a brain capable of an intelligent behaviour based on the increasing penetration of IoT and their growing connectivity? Unlikely. The quantitative difference is way too large.
However, there may be two ways to achieve an emergent intelligence: interconnecting localised artificial intelligence to create a city wide Artificial General Intelligence (AGI) or, and this is what Derrick and myself will be discussing in the December 12 webinar (you can enrol in the free webinar here), by interconnecting citizens, creating a single organism composed of IoTs (taken as sensing and interacting points in a city) and citizens. Notice that a city showing AGI is on a completely different level from a city having, as many smart cities have today, points of localised artificial intelligence. These are making the city smarter, not intelligent. AGI is bringing to the fore a wide variety of issues that are central to philosophical and ethical debates. Our cells are smart but you can get rid of million of them with no ethical or philosophical implications arising. Quite a different story if you are considering the whole organism made up by those cells. These aspects are addressed in the IEEE FDC Symbiotic Autonomous Systems Initiative.
Citizens are sophisticated sensors, they are also biased sensors (they see what they want to see, process data and act according to their individual goals). If they become connected to the city neural soft infrastructure their individual complexity becomes a factor in the overall complexity of the city pushing the whole beyond the minimum complexity thresholds. Notice that you cannot simplify this system without losing its “essence”. It is both a complicated AND a complex system.
More thoughts in the following post.