Driverless cars might be for some the answer to a host of ills, not least of which are dangerous driving and road rage. Indeed, smart robots at the wheel would not bid insistently for certificates of entitlement or reject handling multiple bus routes each week. Yet human traits are also valuable, for example, to read the behaviour of road users accurately. A human would safely cross continuous white lines to get around a stalled vehicle, whereas a robot would not be programmed to break the law and might cause a jam by waiting.

Still, there's merit in tapping technology for repetitive tasks in stable environments. Robots could cover many kilometres smoothly along expressways without rest breaks. Paired with apps, these vehicles could meet shifting demand, optimise sharing and mitigate route congestion. In sufficient numbers, robots could make urban mobility more efficient. But to extend their reach to busy areas, road user discipline will be needed along lanes meant for autonomous vehicles.

Key reliability and safety concerns will no doubt be addressed as the four groups here testing autonomous vehicles make more progress. Even so, public acceptance should not be taken for granted. A survey this year showed that three-quarters of Americans are not ready to embrace driverless cars. While few bat an eyelid when planes are in autopilot mode, they are not ready to trust algorithms to run vehicles, perhaps because they view driving as "a more subjective, personal experience", as a Harvard Business Review commentary posited.

While 90 per cent of crashes in the United States are due to human error, even a single mishap involving a self-driving car is enough to set off warning bells there, it seems. Hence, the human factor must not be overlooked when developing useful technologies. People's perceptions of clever machines play a key part in the social diffusion of inventions.