“…if a ball were to roll onto a road, a human might expect that a child could follow. Artificial intelligence cannot yet provide that level of inferential thinking.”
This quotation, from a 2012 paper put together by KPMG, has already been overtaken by the extraordinary progress in the development of self-driving cars. But programming a self-driving car to anticipate a child following a ball is the easy part of the problem. The tricky bit is programming the car’s response.
“Slam on the brakes”, is the obvious answer. But it is one that ignores the problem that such a programmed response will have invented an exciting new game for children: throw the ball and watch the car stop.
Here is the vision of Google co-founder Sergey Brin, quoted in the New Yorker in 2013:
“… if cars could drive themselves, there would be no need for most people to own them. A fleet of vehicles could operate as a personalized public-transportation system, picking people up and dropping them off independently, waiting at parking lots between calls... Streets would clear, highways shrink, parking lots turn to parkland.”
Absent from this vision are pedestrians and cyclists, who would have to be cleared from busy urban streets for it to become reality. Cars can be programmed. People are more difficult.
All of the descriptions and video demonstrations of progress that I have found so far with the help of Google, itself the leading proponent of self-drive cars, demonstrate quite convincingly that, in a future in which all cars are self-driven, interactions between cars could be controlled in a way that would make car travel safer and more efficient – on motorways or on any other roads from which pedestrians and cyclists are excluded.
But these descriptions and demonstrations also stress that, in the case of interactions between cars and people outside cars (pedestrians and cyclists), the cars will have to be programmed to behave “deferentially”. Moral reckoning to one side, anticipation of the public relations disaster that would follow the first killing of a child by a driverless car demands failsafe programmed deference to those on the street but not in cars.
This deference would clearly become obvious to pedestrians and cyclists and, secure in the knowledge that they were now kings and queens of the road, their behaviour would surely change. Pedestrians would no longer cower at the roadside trying to judge whether gaps in the traffic could see them safely to the other side. They would be liberated to stride confidently into the road, knowing that traffic would stop for them. And all cyclists, not just children, could enjoy the freedom to cycle two or three abreast with friends, holding up middle fingers to the cars honking behind.
So how might this play out in Britain, where cycling is rising in congested city centres such as London, but falling in most of the country? Government policies point in different directions at once.
In his last Budget, chancellor George Osborne appeared to share the Google dream, announcing £100m of government money for the country’s “brilliant automotive industry … to stay ahead in the race to driverless technology”. But this dream will only become a reality if pedestrians and cyclists are removed from our streets.
At the same time the mayor of London is promising to spend almost 10 times that sum over the next decade on cycling. Boris Johnson, like Sergey Brin, also has a vision: one of cleaner, quieter, safer streets with road space reclaimed from cars, and car parks converted to parkland.
However, the mayor’s transformation will be achieved not by driverless cars but by bicycles. The model is the Netherlands, with three London boroughs designated ‘Mini-Hollands’ in the hope that the programme, “will help make them as cycle-friendly as their Dutch equivalents”.
The mayor’s Dutch vision notes that cycling in London has trebled in the last 10 years. Some 24 per cent of vehicles on the road in central London in the morning rush hour are now bicycles; the aspiration is a further doubling by 2020. Johnson’s ambition is to make cycling, “normal, a part of everyday life”. And although his plans include some segregated cycle super-highways, he is clear that “nothing I do will affect cyclists’ freedom to use any road they choose”.
But just as self-drive cars will be designed to stop for children and their balls, so they will stop for pedestrians and cyclists. And if these are allowed to use roads alongside self-drive cars, then drivers will spend most of their time going nowhere. The chancellor’s and the mayor’s visions for our roads are contradictory.
This essay was commissioned on the assumption that I would be sceptical about the idea of driverless cars. But as a London pedestrian and cyclist I am beginning to think that they are a great idea.
John Adams is emeritus professor of geography at University College London.
This article appears in London Essays, a new journal from the Centre for London. Issue 2, looking at technology, is published today. You can read it in full here.