In 2017 the British artist James Bridle produced a video of his work, Autonomous Trap 001. It showed a car park on the edge of a winding road, ascending Mount Parnassus, Greece. A car, which the artist himself has converted through an assembly of sensors and motors into a rudimentary autonomous vehicle, is stationary. The artist opens the trunk and takes out some salt, and draws a white line around the machine, encircling it with road markings. [The salt is used to symbolise the magical power of the marking to conjure the vehicle’s movements]. The car is then turned and soon reveals that it cannot move out of the circle. It is both enchanted and rendered powerless by the markings.

This is not the place to talk about the technology of AV or their viability. So far, there have been numerous accounts of ongoing trials, both successful and sometimes fatal. The investment in AV started within DARPA on behalf of the Defence Department. In 2004 and 2005, they set competitions to encourage developers to use machine learning to create self-driving vehicles. The 2005 contest was won by the Stanford team led by Patrick Thurn, who four years later re-appeared as the head of the Google self-driving car project. The following year, he announced: ‘our goal is to help prevent traffic accidents, free up people’s time and reduce carbon emissions by fundamentally changing car use.’[1]

Since then, software and transport companies are all investing in developing this technology, and governments are starting to legislate for the potential future. Rather than focussing in on the technology itself, we must think of its places within the urban imaginary. What impact will self-driving cars have upon our experience of the platform city?

In Ghost Road, Anthony Townsend predicts that this change will be rapid. All taxis could be self driving by 2030, and by 2050 ‘most human-driven cars will be gone.’ More significantly, he raises the proliferation of different types of autonomous vehicles. We tend to think about cars, but he demands we also consider delivery buggies, bikes, scooters, cars, vans, buses and articulated lorries. The whole system of personal transport as well as logistics and delivery is on the verge of radical transformation.[2]

Most significantly, Townsend asks us to consider in which direction this transformation will take. He has two horizons: Self-driving suburbs, where AV are privatised and we are allowed to travel further in individual pods sealed off from the rest of the city; or there are Car-lite communities where AVs are communal, the car is eradicated, and replaced by micro transport systems for local transit. As a consequence we are living shared lives, closer together.

How does this impact on the experience of everyday life? In his essay ‘The Street as a Platform’ the urban thinker Dan Hill, demands that we need to think this from the ground up. He imagines a city in 2050:

‘Occasionally flocks of ‘Drivers’ swoop through the streets, curving gracefully around the greenery and any other obstacles in their path. Over time, these autonomous vehicles appear to have carved a broad sweep through the foliage; or rather, the greenery has slowly encroached on what was once road space, in a symbiotic relationship with the Drivers. There is no real definition between street and what passes for kerb, and building and flora. There are no traffic lights, fences, street markings, barriers, traffic islands, bollards, drains, road signs, few if any pylons, no step-down transformers, switchboxes. Traffic, from Drivers to bikes to animals, move in all directions at once. Forms ease out of one into the other. . .

There are perhaps three or four drivers every five minutes, each of varying shapes and sizes depending on function. Some carry one or two people, others four or five, or nine or ten. Some drivers carry goods, others are mobile workplaces, bars, medivans or civic service touch-points. There is no parking as far as the eye can see — drivers slide in, pause briefly to pick up, drop off or interact in some other way, and then exit the stage left or right, heading for their next appointment.’[3]

In this scenario, there are no private vehicles anymore, only easily hailed cabs or transit, servicing different numbers of groups. As a result there is no need for parking. The streetscape is liberated from stationary hunks of metal. It is unclear who owns these AVs, is it Uber or the City Hall? Do car manufacturers exist anymore or have they all been amalgamated into one service? Is the locomotive software determined by an ISO, international standard protocol, or are there different protocols for particular vehicles? What happens if a vehicle breaks down? Or even worse, a crash?

The autonomous vehicle, as it is currently being developed by Google, Tesla, Dyson and others, is the solution to a technological problem, but we have not even started to consider the social consequences. Dan Hill attempts to imagine how it might be integrated into a familiar cityscape and make it utterly alien – just as much as the transition from the horse to the SUV did in the twentieth century. As Tim Berners Lee, the man who created the HTML code at the heart of the World Wide Web, said recently: ‘We don’t have a technology problem, we have a social problem.’

In 2017 the German government has released their first response to the most pressing philosophical question concerning the introduction of Autonomous Vehicles into the urban environment: the trolley problem. This is the moral question of what to do if you were driving a vehicle along a cliff road and you suddenly come across a group of children in the road. You cannot avoid the children without driving off the cliff and killing yourself. So what do you do – kill yourself or drive into the children?

As a driver you have the agency to make your own moral decision, and face the consequences, but who makes this decision in an AV? Who should decide whether the algorithm makes one choice or another? Do you presume a company like Volvo will make offer one option – the safety route, say. But Uber offers another – preservation of those who are paying. Tesla autopilot currently has three settings: mild, average and ‘Mad Max’. No joke. And in the end, should we let the market decide this moral quandary?

The German ethical committee decided that self-driving car should choose to hit whichever person it determines it would hurt less, no matter age, race, or gender. That the preservation of life was more important than property. These seem eminently progressive and considered conclusions but it raises some terrifying concerns: If a car can discriminate by gender, race or age, what if, at some point in the future, these moral algorithmic loops can be hacked. Can we encode an ethical framework into the whole networked environment of platform urbanism?

Furthermore when the public realm is in the hands of a plethora of private concerns – platforms, API, app developers – each with their own code and frameworks, how are they going to find an ethical consensus? At present these moral questions are being debated in the US at the state level. California, Nevada and Michigan have all produced different guidelines. But can a car that can cross state lines, and national boundaries, really conform to different ethical standards? Or will the power of the algorithm supersede whatever kind of legislation the local or national executive can impose.

And, therefore, in the scenario of a crash or accident, can we take an algorithm to court? Are we expected to devise new laws based upon the data produced during the extensive testing period, or should we be establishing limits and boundaries from the outset that the technology must conform to?

Finally, we have to note that AVs will be, as Townsend notes, ‘Our first long-term, large scale test case of cohabitation with artificial intelligence, the fast advancing crop of software that seeks to replicate, or at least replace, human cognition.’ For some this is exciting, but as James Bridle’s art reminds us – we have no idea how machines think, we never can; we are dealing with a form of magic that we cannot comprehend – and it does not comprehend us. Call it magic if you like.