Many businesses have come to depend on the car, in a complex, layered ecosystem. At the centre are the manufacturers and their many hundreds of suppliers making parts, and just outside of this is the support infrastructure of showrooms, garages and filling stations. But then there are all kinds of other car-dependent businesses, such as car parks, service stations, and retail parks. The popularisation of car ownership promoted commuting, thereby allowing suburbs to spread away from town centres, and making fortunes for farsighted landowners and developers. Building and maintaining roads and motorways became big business too, while many other industries dependant on the proliferation of horse-drawn transport declined.

This dependency is underpinned by our own behaviour. For many, the car is the among the biggest purchases they ever make. From then on, it influences how and where we live, work and shop. But car ownership is as much about status as transportation, and many of our other status-related behaviours involve the car in some capacity. Perhaps we also think differently too: learning to drive is more about foreseeing danger and developing awareness of other road users as it is about making a vehicle do your bidding. The car forces us to accept responsibility for the welfare of strangers, and has also permitted us to extend the geographic distribution of people with whom we can meet and keep in touch.

Many signs point to the availability of autonomous cars over the next two decades. Google, as well as incumbent manufacturers like Nissan and VW, are known to be investing heavily in their development. Right now, while they are being developed, they look like conventional cars with sensors bolted to the outside and all manner of electronics stacked inside. But there’s no reason for them to stay this way. Cars that don’t need drivers also don’t need driver controls or any the other practicalities of driving. As most accidents are attributed to human error, it’s possible the extensive safety features of contemporary cars will also be redundant. This means that the driverless car prompts a complete rethink in car design. Whereas today cars are designed from the perspective of the driver experience, autonomous cars could start with the passengers. This will likely affect layout, but may also allow them to be made lighter and therefore cheaper to build, ship and run.

As the cost of manufacture drops, we could see new entrants to the market; perhaps disrupting the symbiosis of makers and suppliers. A company like Bosch makes a wide range of car parts, and also has divisions making common appliances—-if the cost of development were to fall, it would be well-placed to rival incumbent car brands. As with the mobile phone market, we may see a new symbiosis between hardware and software manufacturers.

With everyone a passenger, the focus of car marketing will shift away from the driving experience. Performance will be downplayed for other features, similar to how we have been prepared to sacrifice battery life for more functionality from our smartphones. The experience of using a motor vehicle will shift towards other passenger transport, allowing passengers to eat, work or simply relax while on the move. Perhaps also it will become less desirable or practical, maybe even feasible, to own a car outright. Parking in highly populated areas is at a premium, but there is no need for autonomous cars to be parked near their passengers. Having dropped them off, the driverless car could scurry away to find parking much further away if need be. But it might prove more useful for that car to take up the service of other potential passengers in the vicinity, much as taxis do now. This suggests buying the use of a car (as opposed to outright ownership) may prove more desirable for many, particularly in built-up areas. This model, akin to Spotify versus the CD, could be highly disruptive, not only to the business models of taxi drivers and hire companies, but also to car makers and their dealers. It may also prompt many other car-dependent businesses, such as owners of out-of-town retail parks, to consider their involvement in the economics of car usage—-perhaps subsidising the cost of transport to their sites.

The biggest impact of autonomous cars would be on the transport infrastructure. Being able to drive nose-to-tail, in narrower lanes, and with a reduced need for parking, many more driverless cars could fit into the existing road-space. But this infrastructure has an ecosystem of its own, and employs many people. From the recovery services to sign-makers, and not forgetting traffic wardens, many lines of revenue could dry up as they did for blacksmiths and farriers a century ago. The impact may stretch further than the roads: an autonomous vehicle designed for long trips could take the form of a hotel room on wheels, thereby rivalling train operators, airlines and even the hospitality industry. But perhaps the many losers could diversify into reviving country pubs, or the further expansion of suburbs.

As the technology advances, many unknowns remain. Two of the most significant are particularly tricky. First, even though autonomous cars reposition the role of the driver from human to machine, they share characteristics. In an adverse situation, a driverless car will slam on the breaks—-perhaps faster and harder than a human might—-but in any such occurrence there remains an element of chance. While autonomous cars stand to reduce accidents resulting from human error, some accidents remain unavoidable. For such instances, a new legal framework would be needed to determine liability. Are manufacturers prepared (and can they afford) to extend their warranties this far? An alternative is to construct an entirely separate and isolated infrastructure for autonomous cars, but such projects are wildly expensive and would also be starting from a long way behind the current road networks.

Secondly, a sudden switch to driverless cars is a big ask. The shift from horses to engines was relatively subtle: the horse was a power source and little else, making it much easier to swap out for new technology than the many functions of a driver. It is more likely the cultural and economic shift to autonomy will take much longer than the development of a production-ready driverless car. Despite landing on the moon in the Sixties, we don’t yet live there… although the space race prompted the development of many technologies we use daily here on earth. Similarly, the learnings from the development of autonomous cars may, in the medium term, manifest themselves as further-improved augmentation for (human) drivers. Many of the technologies used to make cars autonomous, such as their many sensors, are available today on production models, so this journey is already underway.

Perhaps it is a more feasible starting point for the autonomous abilities of cars to be used in specific areas, where the most benefit will be realised. A city might designate a no-driver zone in its centre, for example, where parking is rescinded but autonomous vehicles are free to come and go. This would tackle congestion, pedestrian safety and pollution, but the cars would likely retain the controls and safety features of their conventional brethren, and thereby many of the most attractive possibilities of truly autonomous cars would be lost.

If totally driverless cars is the technological destination, many of the benefits come from allowing these vehicles to share or adopt the existing road infrastructure. In most places, a sudden switch would be difficult (and expensive) to orchestrate, so it seems more likely the technology will instead creep in. However, as driver-assisting technologies develop further, the roles of driver and passenger become increasingly blurred. With a mix of cars at different stages in this evolution on the road together, at what point is it appropriate for a driver to pay less attention to the surroundings? And while there’s still a human driver onboard, who is accountable for the decisions made by the machine?