Artificial intelligence and the city - a parking story
When cities implement artificial intelligence without involving the population, it can rapidly backfire. The case of Brussels and smart scan-cars illustrates it.
Priest versus robot
Modern times often lead to unexpected situations. Belgium, sporadically appearing in the news for its puzzling way of finding solutions, provides a natural fertile ground for such stories.
A few months ago, a Belgian judge had to settle a case between a priest and a robot.
The problem? The priest was fined by a scan-car for parking without any ticket as he was posting a letter in a mail box across the street.
The priest argued that he had to right to stop because it was only a few seconds but the scan-cars, detecting the vehicle being empty and without a valid ticket, had imposed the fine.
The holy man then tried to contest the ticket to the company owning the scan-cars, but, to make matter worse, only received automating answers from a computer.
He eventually brought the case to justice, happy to be finally dealing with a human.
Scan-cars in Belgium
It is not the first time that scan-cars are making the headlines in Belgium. Brought onto the streets for the first time in 2019, more and more cities in the country are using those machines to handle parking tickets, including the capital city Brussels.
Circulating all day long, continuously photographing the cars parked and automatically analysing the images, scan-cars are expected to bring efficiency and automatisation in parking management, as clearly stated on the website of the city of Brussels.
With such intelligent cars, municipalities are now able to control 1,200 cars per hour, versus 50 for human agents, i.e. 20 times more. In parallel, the system automatically send the fine to the home addresses of the offenders, offering a seamless tracking of the fine.
Thanks to those improvements, some communes have been able to increase their revenues from parking tickets payments by 50%.
Yet, the system is far from being perfect: around 27% of fines had to be paid back, according to the opposition party.
The problem stems from the precision of the system, or rather, the lack of thereof.
For one thing, the scan-cars cannot detect handicap cards and other exemption permits, such as card authorising employees to park near their place of work.
Furthermore, they cannot make the difference between a car parked and a car stopped on the street, like the Belgian priest, nor understand if the driver is on the way to buy the needed parking tickets or even in the car.
While the gains looks simple and clear, the costs and impacts of scan-cars are unsure and seem to have been overlooked by the authorities. Was it worth it?
The tree hiding the forest
This example highlights the problems that can arise when using technological solutions in the public sphere.
One of them lays in the way problems and solutions are defined in such projects.
Implementing technology typically aims at optimising one metric, such as traffic flows, building capacity or energy consumption. Models and technics are then chosen and developed to improve this target metric and provide a system with the highest efficiency.
“What gets measured gets managed” as the maxim goes, but whatever is not taken into account in the target metric will not be dealt with, leading to unexpected outcomes.
In the case of scan cars, the target metric is clearly the number of cars controlled. This is not a bad metric per se: as they put it, better control will lead to better usage of the parking use, i.e. an optimisation of the parking space of the city.
Yet behind this simple metric stand broader and arguably more important questions, such as accessibility or quality of life, and the impact on those is not clear.
This leads us to a second problem of using technology: simplifying the problem with a metric does not only hide the impacts, it also dodges the needed debate about what the city wants to achieve.
Accessibility or standard of living are political decision that have to be compromised, not optimised, and using a single metric can make us look away from the underlying issues.
Those metrics are not coming out of nowhere. It is easier for companies to sell products and services based on powerful arguments, cost efficiency is often a strong one.
It is not a coincidence that the city of Brussels is claiming the number of cars controlled as a benefit of the scan-cars. One of the selling argument of the companies leasing the scan-cars is that cities will be able to get back the money spend on those cars with the subsequent revenues in ticket.
Talking about efficiency might work well in the private world, after all the buyer knows what he is buying (most of the time), but in the public realm, what means efficiency for the company might not mean the same for the end user, the citizen.
Yet, addressing the needs of the residents is more complex than talking about efficiency, but that is what urban governance is about. As efficient as technology can be, we should not ignore the impact it can have nor the difficult issues it needs to resolve.
Let’s talk about it
While Belgian laws obliges an approval from municipalities to operate scan-cars, this is not always the case and clearly not enough.
Scan cars is just one example of public entities struggling to reconcile efficiency and legitimacy when implementing technology.
From regulating traffic in Brussels to the choice of schools in France or the risk assessment for criminal offender in the United States, public entities often turn to technology amidst a urge to optimise the usage of limited public resources such as roads, schools or prisons.
Just like scan cars, those solution have been criticised for their lack of transparency whether about the data they collect or the way they are making decision. Education, justice or public space are societal topics, that needs to be discussed openly, not reduced by a metric. Failing to check the data collection and development can lead to unexpected outcome like the scan cars in Brussels.
One way to ensure that the right metrics are taken into account and that the question is actually debated by the public is enhancing transparency around data collection and algorithms.
The city of Amsterdam for instance, is hosting an Algorithm Register, a website gathering information about all the algorithms used by the city, from what data they used to how it was set up.
One of them is actually about the algorithm used by scan cars for parking control. On their page, you have access to how it is developed, its risk management safeguards and contact information if you have further questions. Although they are not precisely showing how their algorithm works and how they are making sure the right cars are fined, the website makes it possible to have a discussion about it.
Another way to do it is to require a mandatory review of the algorithm by a dedicated board. In some US cities, such as New York or Santa Clara, it is mandatory to go through an audit process and impact assessment for certain type of algorithms.
Finally, procurement process should become more dynamic so as to allow continuous discussion between companies and city. For instance, the city of Barcelona has updated its procurement process for ICT product to center procurement around the problem it wants to solve, rather than on very detailed technical lists, making technological projects more adaptable.
From efficiency to urbanity
All of those improvements are trying to bring technology closer to the city. To measure the true impact of technology and to foster debates around it, we should switch from making city more efficient to making technology more urban. As developed by Professor Saskia Sassen, this means to make technology attentive of the local specificities of each city and more compatible with their dynamic forces, in opposition with the scalable 1 size fit all mantra of today’s tech industry.
This might be the only way to make the technological transplant stick to the city and to avoid its reject by citizens, just like our Belgian priest rejected the scan-cars decision.
Thankfully, his devotion ended up being rewarded as the fine was finally cancelled by a judge a few months ago. While efficiency is the prerogative of technology, resilience and perseverance seems to remain the gift of the city.
- Green, B. (2019). The smart enough city : putting technology in its place to reclaim our urban future. MIT Press. https://mitpress.mit.edu/books/smart-enough-city