In the popular imagination, the rise of self-driving cars will mean greener cities, safer roads, and happier workers. With technology, speeding, braking, and reaccelerating-which are responsible for much of the gas burned on roads, today and the resulting air pollution-would be a memory. But despite these opportunities, there are certain reasons why driver less vehicles may fail to live up to some of these expectations, or at least take a little while longer to achieve them.

    First, truly automated technology - at least the kind that requires no human involvement -is rare in every sector of the economy, and there's little reason to think driving will be any different. There is a permanent need for a human backup in case driving, that means human drivers will still bear the final responsibility for the cars they're riding in.

    A related challenge-particularly for driverless cars-is the complexity of the systems involved. For example, some cars today require over 100 million lines of software code (代码) to keep them running More code means more functions, which, to some degree, also means more opportunities for software errors. Driverless cars raise a challenge in this regard and possibly cause more opportunities for errors- errors that can lead to accidents.

    Yet the greatest challenge facing driverless cars isn't technological, but psychological. Surveys consistently reveal global skepticism about the technology. Americans, for example, remain more worried than excited by driverless cars. Over half of the American public says it would not ride one given the opportunity. This shows a deep-seated lack of trust in the technology. Similar ideas have been expressed worldwide.

    What's more likely is that human drivers will continue applying some control over their cars, regardless of how automated those cars are. Some may quarrel with this idea. But automakers have long been preparing for it. Just ask Nissan, Waymo, and Zoo, all of which are exploring ways for human drivers to retake control of a car should automation fail. In other words, driverless does not-and will not-mean humanless.