Since early 2016, driverless cars have been tested in city centres across the UK, in London, Milton Keynes, Bristol, and earlier this month, in Coventry. One of the stated objectives of these publicly funded trials is "to understand how people respond to, engage with and accept automated vehicles," as the project website of one of the projects, the Greenwich Gateway, puts it. However, while the ostensible aim is to test whether driverless cars are able to operate on public roads, and to share this space with other road users, it is clear on which terms the trial organisers expect this peaceful co-existence to be realized: the car's. In Greenwich, cyclists have been banned from the urban paths where driverless tests take place. An organisation called the Pedestrian Liberation Front is considering to lodge a formal complaint against an Ocado trial with driverless delivery vans in Woolwich: it alleges that pedestrians are discriminated against by the trials as they take over streets previously marked as “pedestrian-friendly.”
To justify the testing of driverless cars in the street rather than in the laboratory, it is often argued that this is the only way in which the capacities of these machines to interact with other road users, amidst social complexity, can be examined and developed. But if we consider the design of the tests in more detail, it is clear their purpose is also to contain 'social complexity' and to render it manageable for cars. When self-driving vehicles were tested in the city centres of Milton Keynes and Coventry, the press releases announcing the tests were embargoed until after they were concluded. By the time the trials were made public, the driverless cars had disappeared from the road. The justification provided for these restrictions on publicity is ‘safety’. In an exchange with one of the organisers of the Coventry trial, it was suggested to me that if the public would know in advance when the cars would appear in the streets, it would get too dangerous: people might show up in large numbers, or play tricks on the vehicles.
So, we are in a situation in which technical experiments are conducted on public roads, with public funding, without the public being informed beforehand, let alone giving consent. It is unclear whether driverless street trials have ever received formal, regulatory approval from a public body in the UK. The tests are conducted in accordance with a government-approved “Code of Practice” but this code endorses a “non-regulatory approach” to the introduction of driverless cars to society: as the UK did not sign the Vienna Convention on Road Traffic, it is apparently the only country in the world where drivers are allowed to take their hands of the wheel without prior approval from state authorities. When it comes to the public legitimacy of these experiments in society, the current thinking seems to be that as long as the engineers can get these cars to work, public approval will follow. As a representative of the Milton Keynes Transport Catapult put it: “If people can see that these vehicles are capable of driving themselves, they can gain trust and make sure that all the correct safety measures are in place for these cars to drive themselves, and then that is exactly what we should be doing.”
However, while we may question the legitimacy of these trials, it is also important to probe the implications for public knowledge: is it really possible to investigate the capacity of machines to co-exist with humans under these conditions? As cyclists are banned in Greenwich, and citizens are not informed of the trials taking place beforehand, these other road users aren't really able to participate in the experiment. From their perspective, driverless street tests do not qualify as an experiment in co-existence between human and machine, but rather present an exercise in displacement.
Also more generally speaking, I don’t think these driverless street trials, in their current design, can be characterized as experiments from the point of view of society. For engineers, these trial are a way for driverless vehicles to learn how to get along with city transport. But, from a social perspective, very little can be put to the test in these trials. Given how little citizens and social actors know, and can do, it is far more accurate to say that these trials are demos, or show trials, not tests. They fail to examine a central challenge of co-existence, one to which social researchers of smart technology, like Lucy Suchman, have been calling attention for decades, namely the challenge of how to exist alongside beings who are different from us. How to establish mutual relations across chasms that separate machines from humans, engineers from the public?