Driverless cars are coming fast, but perhaps not in the way you expect. Here are some common misconceptions, clarifications and questions that remain.

Driverless isn’t the same as autonomous

An autonomous car has a driver, but it has technology to let that driver choose not to drive for a period of time. The driver remains at the wheel and at any time can choose to take control again. Autonomous cars may also be programmed to alert the driver and return control to him or her in certain situations: when a difficult choice is required, for example, or when the car is leaving a specified zone where autonomous driving is permitted. A driverless car has no human driver controlling it from the inside. Programming or external control is responsible at all times for navigation.

Autonomous cars are on the market now; driverless cars will be within five years

Major car companies are already selling autonomous cars on the mass market. In Japan, Nissan is selling a mini-van that offers, as an option, the capacity to drive autonomously on a specially reserved lane on certain highways. Most consumers who buy this mini-van are choosing to pay extra for the autonomous option, according to Carlos Ghosn, Chairman and Chief Executive Officer, Renault-Nissan Alliance, France. “Consumers prefer to have more comfort and less stress,” he says. He predicts that by 2018 autonomous cars will be able to operate throughout multi-lane highways, and that by 2021 or 2022 we’ll see mass market sales of truly driverless cars.

Technology isn’t an obstacle for driverless cars

Prototype driverless cars are already working in controlled environments. Even if the technology isn’t yet installed for driverless cars on a mass-market scale, it won’t be a significant challenge. The main requirement is an extremely reliable 5G wireless network so cars can connect with each other seamlessly and close to instantaneously. This connectivity is needed so cars can not only sense their surroundings but also communicate to avoid accidents and optimize the driving experience. “A smart phone might be able to do a lot of this connectivity for cars,” Paul E. Jacobs, Executive Chairman, Qualcomm, USA, says. “That will help drive down costs.” He predicts that this technology will probably be ready for widespread use by 2020.

The human-machine interaction point is where risk enters

Autonomous and self-driving cars can be programmed to create a safe environment, far safer than the current transport scenario, in which more than 1.2 million people a year die in car accidents. But what happens now when an autonomous car hands control back to a driver, who may have been napping or watching a movie? And since for many years most cars will have human drivers in control, what happens if non-autonomous cars (or pedestrians) fail to act rationally or according to probabilities? It is because of these friction points that the initial roll-out of autonomous cars is taking place in a controlled environment – a dedicated highway lane – and that self-driving cars too may first enter controlled spaces such as urban centres or corporate campuses.

Society still has a lot of questions to answer

One moral question has already become famous: what should a self-driving car do if it must choose between saving the life of its driver, or saving the life of a greater number of people in a different car? Most people answer that the car should choose to save the greatest number of people, but they also probably wouldn’t ride in a car that would sacrifice their lives to save others. Regulators and industry groups will have to make tough decisions about accountability: who will pay the price when accidents happen and no one is at the wheel? Other questions to be answered include cybersecurity – how to keep this autonomous driving network safe from hackers – and the unemployment that driverless cars may create. The technology will create new jobs too, but it’s far from clear that displaced workers will be ready for them. That may further fuel populist political tensions.