This article is about how the driverless revolution could be trickier than we think. The article discusses the results of Arthur D. Little’s study ‘Capacity effect of autonomous vehicles’ which points to a fundamental difficulty: driverless cars are programmed to obey the rules and not take risks, for example when changing lanes. The study suggests that this could increase traffic jams by more than 16 per cent. Then there is the challenge posed by human behaviour — no company would ever programme a car to run over civilians stepping into the road. This means pedestrians and cyclists will be able to run rings around them. On top of that, the study suggests that the availability of self-driving vehicles could result in fewer people taking public transport, resulting in a 50 per cent increase in overall car use by the middle of the century.