Autonomous vehicles are currently subject to a lot of hope, hype and hysteria, especially after the recent fatality in Arizona, when a pedestrian was struck and killed by one of Uber’s self-driving cars. Autonomous vehicles will likely have as significant an impact on our lives as electricity, computers and the internet all did. However, just like the aforementioned innovations, we cannot predict all the varied impacts and implications, both positive and negative, of self-driving vehicles. Here, we’ll offer a brief overview of what we do know and what’s still up in the air.

What We Know

Autonomous vehicles are coming.

In fact, they’re already here. Autonomous buses and shuttles are currently being deployed in city centers , sprawling work sites, and airports; driverless trucks are delivering beer long distances; and Waymo’s test set of self-driving cars have clocked five millions miles on public roads. Even autonomous flying taxis seem to be in our near future. It’s no longer a matter of if, but when.

Autonomous vehicles will change the fabric of a city.

In a future where fully autonomous vehicles are the norm, there will be no need for lanes or road signs or stoplights. Cars will be connected by high-speed internet and able to communicate with one other, their occupants, as well as any central or local coordinating infrastructure. Getting lost will be a thing of the past, thanks to lightning-fast access to the latest, most accurate maps and finer resolution GPS (coming soon). Road furniture, as we know it today, will be obsolete.

Thanks to more precise and safer driving, theoretically, roads could become even narrower, allowing for more of the city to be given back to the community, pedestrians, and other modes of transport. Urban life as we know it, will change drastically.

The flow of traffic will be smoother.

If vehicles are explicitly communicating with each other, or with a remote coordinator, they can efficiently accelerate or decelerate so that stopping becomes altogether unnecessary at intersections — assuming, of course, that roads don’t all become sets of “Boring” underground tunnels). With this logic, compared to an intersection with similar density of traffic and human-manned vehicles, traffic flow should be significantly more seamless and efficient in a world run by autonomous vehicles. In addition, if vehicles are sharing explicit intent with each other, the others on the road can adjust and plan accordingly. Vehicles can change lanes or speed gently, surprising no one, leading to better safety and a smoother, more fuel efficient ride.

Automotive accidents will decrease significantly.

In 2016, over 37,000 people died in U.S. motor vehicle accidents. Globally, around 1.3 million people die in crashes each year, or almost 3,300 deaths per day, with an additional 20–50 million injured. These numbers will decrease significantly with the adoption of autonomous vehicles, due to explicit communication between local vehicles, for one. They will know exactly where neighboring vehicles are headed, their velocity, and what their next moves are. Another factor is that these driverless vehicles will be risk averse and constantly on the lookout for potential hazards using their suite of onboard sensors such as LiDARs, cameras, and radar.

People will still die from automotive accidents.

Recently, Elaine Herzberg died after being struck by one of Uber’s self-driving vehicles in Tempe, AZ. , Unfortunately, she won’t be the last. People will die as we develop autonomous vehicle capabilities — just as people have died during the development of aircraft or space or submarine travel. Major innovation always has a human cost, but driverless technology should result in a significant reduction of annual automotive deaths.

Autonomous vehicles will change the nature of work and leisure.

Driverless vehicles won’t just be a mechanism to get from point A to B. A commuter who no longer has to manually drive to work can instead use that time typing away on a computer, , which could lead to increased productivity or a shorter actual workday leading to a longer workday. Parents who don’t need to watch the road can instead use that time to play a board game with their kids or help with homework. An amorous couple could Netflix and chill. (For extreme commuters, the car could literally be the bedroom, as I’ll discuss further below.) Our perception of and relationship to a vehicle’s interior, as a spatial component of our life, will change.

Just as smartphones have blurred work and home life, autonomous vehicles too will blur home, transport, and office life.

What We Don’t Know

When will we be fully autonomous?

If we’re being honest, we don’t know the answer to this question. While we might expect autonomous vehicles to be more commonplace in 10 years time, it probably won’t be until around 2050 until full “Level 5” autonomy. Why 2050? That’s how long it took for us to transition from horse to car.

State regulators have been passing legislation to support testing of autonomous vehicles faster than many others have expected, including green lighting driverless testing in California this past February. Undoubtedly, there will be some set backs in reaction to the Tempe fatality — and Arizona’s governor has suspended Uber’s testing capabilities — but it goes to show that many potential hurdles still exist. All in all, it’s tough to make predictions about the future.

How will we transition?

Between now and 2050, there will be a mix of Level 0 to Level 5 vehicles tested and operated, until we collectively reach Level 5. The U.S. is a car-loving culture, and there will be associated resistance. How this transition will work — considering auto-senescence, planned obsolescence, incentives for manufacturers, and government regulation — is unclear.

It could also be a mess with human drivers likely gaming the system. If they know an autonomous vehicle behavior, like that the car is extremely risk averse and will make way or completely stop in the face of aggressive driving, human drivers could bully driverless cars on the road. Conversely, one study showed that the presence of even some autonomous vehicles improves flow in mixed traffic.

Who will own the vehicles?

In New York, relatively few people own a car, because of dense public transport and readily available taxis, car services, and other shared vehicles. When we are fully autonomous, it’s likely that there will be a large fleet of shared vehicles in metropolitan areas. Thus, the population who don’t own cars, and area they live in, will both be larger than they are today. In rural areas, the network effects from a shared fleet are less likely to take hold and vehicle ownership rates are likely to be greater. In addition, the one percent, or those that require additional privacy or security, will own and run their vehicles. However, we still have yet to fully understand what these patterns will be, and who will own and run these types of fleets.

Who will be the key players?

While Waymo, which is controlled by Google’s parent company, currently has a first-mover advantage in terms of the quality and maturity of their technology, they are by no means assured to maintain their lead over the coming decades. Most major car manufacturers, supported by dozens of OEMs developing component technologies, have already jumped into the fray. It’s perfectly likely that the autonomous vehicle industry will be dominated by a company that has not yet been conceived of or created.

How will traffic density change?

There is a theory that a fleet of shared vehicles who can drop off their passengers and park themselves away from a city’s downtown area will eliminate the local traffic circling for parking. (The current density of downtown parking-seeking traffic is significant, up to sixty percent of vehicles!) However, the counter-argument is that empty vehicles on the roads heading to offsite parking rather than downtown garages and street parking will increase local traffic density. If a fleet were optimally utilized, it could be physically smaller and carry a smaller environmental footprint. If sub-optimally used, including zipping off to out-of-down town lots rather than picking up passengers close to their destinations, they would need to be physically larger. Finally, in very dense urban settings, while intersections, in theory, should allow for a continuous flow of vehicles, this may not be possible in some situations due to excessive density — unless we create more dimensions, either above or below ground. All told, we don’t yet know the net result: a smaller set of efficiently used vehicles, or a larger set of less efficient vehicles continuing to clog the streets.

Who will be liable?

We’ve established that accidents will still happen, and people will be injured and/or die. However, who will be responsible? The fleet owner, the vehicle manufacturer or perhaps the software engineers who program the technology? There are still big, thorny legal issues yet to be decided.

Inevitably, vehicles will run into the “trolley problem,” an ethical dilemma where the vehicle has to decide between two or more actions, each of which carries some cost — for instance, which is worse: swerving left, killing four grandmothers, or swerving right and killing a mother and her baby? Someone has to program those behaviors, or develop an AI that learns to make that decision itself. We aren’t sure who will make these ethical decisions and who will regulate them.

How will this change commuting patterns?

We don’t yet know how autonomous vehicles will affect where people will live and where they will work. In New York, I’ve known several people who commuted 90–120 minutes (each way!) to work. Surprisingly, three percent of American adults are extreme commuters, defined as someone who travels more than 90 minutes to work, each way. With the adoption of driverless vehicles, this can only expected to get worse. If the car doesn’t require a human operator, the passenger can be sleeping. Moreover, the ride is expected to be smoother, and without a need for interior vehicle controls, there is more configurable space, potentially for a bed. Perhaps commuters will get into bed in their car in Phoenix at midnight to wake up close to work in Los Angeles for breakfast. The car could even travel more slowly or take backroads to give the occupant more time to sleep. The scenario seems dystopian, but it’s possible.

How will this affect family life?

If dad is on a 400-mile commute, he won’t be at breakfast with his kids or home in time for dinner. If driverless cars are sufficiently safe and reliable, soccer moms won’t need to truck kids to and from games. The vehicle can take Alice to dance and Bob to school, meaning less time together as a family. Personally, I would take more fun family weekends away if I didn’t have to endure the long drive each way. A family could watch an in-car movie, sleep, and then wake up together in the mountains or at the beach or at grandma’s house. Autonomous vehicles have the potential to impact family life both positively and negatively but we don’t precisely know which way the pendulum will swing.

All in all, I’m excited by the future of autonomous vehicles. I‘ve experienced the birth and adolescence of mass consumer computers and the internet, and this feels like we are on the cusp of another massive societal change, one that will shape both mine and my kids’ lives in unimaginable ways.