It's gonna be weird. Until the day when even people who grew up driving can't rememberwhat it was like to see human drivers. JL
The New York Times reports:
Call it 20%: the day, if it ever arrives, that one out of five vehicles you encounter will be driving itself. Coexistence between human and robot drivers will be
uneasy at best; road rage against the machine will boil over. But
that is the point when it will
be clear that a self-driving future is possible. When car
companies will rip out the steering wheels, designing them in entirely different ways. When hypercommuters will trust their cars to drive them to the office while they sleep. When you risk putting
your child in one, alone.
Technically speaking, self-driving cars are already here. But in even the most optimistic scenario — given their persistent difficulties with situations as simple as a snowy day — robot drivers will remain wildly outnumbered by human drivers for the foreseeable future. Autonomous cars can’t change the world until they’re a major presence on the road.
Call it 20 percent: the day, if it ever arrives, that one out of five vehicles you encounter will be driving itself. In some ways, the 20 percent era is harder to imagine than the fully automated era, because this supposedly unstoppable and frictionless technology will still be embroiled in a total mess. The best, safest models will most likely come with the highest price tags, out of reach for those with average incomes. Coexistence between human and robot drivers will be uneasy at best; road rage against the machine will boil over. One catastrophic accident could still imperil the whole experiment, no matter what the overall statistics show.
But that inevitable mess is precisely what makes the one-in-five future so interesting to ponder. It’s the first point, give or take, when it will be clear that a true self-driving future is even possible. When car companies will actually be able to rip out the steering wheels from some models, designing them in entirely different ways. When big trucking firms and car services like Uber will have begun to remove drivers from their vehicles. When hypercommuters will trust their cars to drive them three hours to the office while they sleep. When you might risk putting your child in one, alone.
Will We Need A License To Ride?
By Tom Vanderbilt
In 2008, I rode in a Stanford University VW Passat called Junior on a closed course in New York. Like a lot of beginners, it moved haltingly, with great caution; its minders in the front seats kept a close watch. But the moment felt transformative. Driving, burned since adolescence into my muscle memory and sensory apparatus, was being executed without my (or any human) involvement. I felt like a ghost in the machine.
The autonomous vehicle went from test-track prototype to showroom reality far faster than I envisioned, propelled as much by advances in sensors and A.I. as by a collective hype cycle that seemed intent on willing these devices into existence, whether society was ready for them or not. Sitting with that “beginner” driver in 2008, as engineers closely observed its movements, took me back to that life-defining moment of getting a driver’s license. In my wallet, sparse with a library card and a school-lunch ticket, the license had real heft. It represented a primordial entry point to adulthood, a vast expansion of social possibility.
As a rite, the driver’s exam — I have failed it in two countries — is a ritual by turns (three-point, of course!) tedious and terrifying. If you can summon nothing else of the 1988 teen comedy “License to Drive,” you can still see James Avery’s gleefully malevolent D.M.V. examiner, wielding his famous coffee cup. Once autonomous technology begins to really take off, will the driver’s license, and the pedagogy behind it, go the way of the stick shift? Will the ability to drive begin to seem some exotic, rarely necessary skill — like starting a fire with a flint?
After all, in a vehicle that meets the Level 5 automation criteria of SAE International — “full-time performance by an Automated Driving System for all aspects of the dynamic driving task under all roadway and environmental conditions” — there would seem to be little point in possessing this vestigial skill. We don’t require people entering elevators, riding in airport trams or letting Roombas loose upon their living rooms to be able to take over operation in case of machine failure; it’s hard to see why our interaction with this automated technology should be any different. Our learning process could be as quick and dirty as it is with other high-tech products: Toss out the instruction manual and watch someone do it on YouTube.
But this possibility, driver made passenger, broaches new territory in jurisprudence. With no licenses, what is the legal basis for restricting anyone — tweens, the elderly — from being ferried about? Young farm kids’ driving the family truck seems as obsolete as tail fins, but what’s stopping kids from being driven? Could children, often more adept at handling the smartphones of their parents than their owners are, have an advantage on older drivers less accustomed to technology? Yet the same fears that keep parents from letting their children walk to school may prevail over the allure of their autonomous vehicle handling the school run. It seems hard to say whether autonomous vehicles would hasten or delay the passage to adulthood.
And as drivers interact with semiautonomous vehicles in the long run-up to Level 5, driver education and licensing, far from becoming obsolete, may become more important, argues John D. Lee, a professor of engineering at the University of Wisconsin-Madison. “Automation has a strong tendency to surprise people with unexpected behavior,” he says. Knowing what your car is doing — and when to take over — could be more vital than merely knowing how to drive. That water on the road — is it a surface puddle, or a veritable lake filling a gigantic pothole? That truck ahead with the dangerously swaying load — do you speed up to pass it? Is the pedestrian by the marked crosswalk waiting for you to yield or simply checking his iPhone? “Driving and managing the automation that is helping you drive,” Lee says, “are two quite different skill sets. Automation-management skills need to be learned as much as driving skills.”
Tom Vanderbilt is the author of “You May Also Like” and “Traffic.”
Empty-Car Surfing
By Rollo Romig
Illustration by Erik Carter
Driverless cars will sometimes need to
Technically speaking, self-driving cars are already here. But in even the most optimistic scenario — given their persistent difficulties with situations as simple as a snowy day — robot drivers will remain wildly outnumbered by human drivers for the foreseeable future. Autonomous cars can’t change the world until they’re a major presence on the road.
Call it 20 percent: the day, if it ever arrives, that one out of five vehicles you encounter will be driving itself. In some ways, the 20 percent era is harder to imagine than the fully automated era, because this supposedly unstoppable and frictionless technology will still be embroiled in a total mess. The best, safest models will most likely come with the highest price tags, out of reach for those with average incomes. Coexistence between human and robot drivers will be uneasy at best; road rage against the machine will boil over. One catastrophic accident could still imperil the whole experiment, no matter what the overall statistics show.
But that inevitable mess is precisely what makes the one-in-five future so interesting to ponder. It’s the first point, give or take, when it will be clear that a true self-driving future is even possible. When car companies will actually be able to rip out the steering wheels from some models, designing them in entirely different ways. When big trucking firms and car services like Uber will have begun to remove drivers from their vehicles. When hypercommuters will trust their cars to drive them three hours to the office while they sleep. When you might risk putting your child in one, alone.
Photo illustration by Tamara Shopsin
Will We Need A License To Ride?
By Tom Vanderbilt
In 2008, I rode in a Stanford University VW Passat called Junior on a closed course in New York. Like a lot of beginners, it moved haltingly, with great caution; its minders in the front seats kept a close watch. But the moment felt transformative. Driving, burned since adolescence into my muscle memory and sensory apparatus, was being executed without my (or any human) involvement. I felt like a ghost in the machine.
The autonomous vehicle went from test-track prototype to showroom reality far faster than I envisioned, propelled as much by advances in sensors and A.I. as by a collective hype cycle that seemed intent on willing these devices into existence, whether society was ready for them or not. Sitting with that “beginner” driver in 2008, as engineers closely observed its movements, took me back to that life-defining moment of getting a driver’s license. In my wallet, sparse with a library card and a school-lunch ticket, the license had real heft. It represented a primordial entry point to adulthood, a vast expansion of social possibility.
As a rite, the driver’s exam — I have failed it in two countries — is a ritual by turns (three-point, of course!) tedious and terrifying. If you can summon nothing else of the 1988 teen comedy “License to Drive,” you can still see James Avery’s gleefully malevolent D.M.V. examiner, wielding his famous coffee cup. Once autonomous technology begins to really take off, will the driver’s license, and the pedagogy behind it, go the way of the stick shift? Will the ability to drive begin to seem some exotic, rarely necessary skill — like starting a fire with a flint?
After all, in a vehicle that meets the Level 5 automation criteria of SAE International — “full-time performance by an Automated Driving System for all aspects of the dynamic driving task under all roadway and environmental conditions” — there would seem to be little point in possessing this vestigial skill. We don’t require people entering elevators, riding in airport trams or letting Roombas loose upon their living rooms to be able to take over operation in case of machine failure; it’s hard to see why our interaction with this automated technology should be any different. Our learning process could be as quick and dirty as it is with other high-tech products: Toss out the instruction manual and watch someone do it on YouTube.
But this possibility, driver made passenger, broaches new territory in jurisprudence. With no licenses, what is the legal basis for restricting anyone — tweens, the elderly — from being ferried about? Young farm kids’ driving the family truck seems as obsolete as tail fins, but what’s stopping kids from being driven? Could children, often more adept at handling the smartphones of their parents than their owners are, have an advantage on older drivers less accustomed to technology? Yet the same fears that keep parents from letting their children walk to school may prevail over the allure of their autonomous vehicle handling the school run. It seems hard to say whether autonomous vehicles would hasten or delay the passage to adulthood.
And as drivers interact with semiautonomous vehicles in the long run-up to Level 5, driver education and licensing, far from becoming obsolete, may become more important, argues John D. Lee, a professor of engineering at the University of Wisconsin-Madison. “Automation has a strong tendency to surprise people with unexpected behavior,” he says. Knowing what your car is doing — and when to take over — could be more vital than merely knowing how to drive. That water on the road — is it a surface puddle, or a veritable lake filling a gigantic pothole? That truck ahead with the dangerously swaying load — do you speed up to pass it? Is the pedestrian by the marked crosswalk waiting for you to yield or simply checking his iPhone? “Driving and managing the automation that is helping you drive,” Lee says, “are two quite different skill sets. Automation-management skills need to be learned as much as driving skills.”
Tom Vanderbilt is the author of “You May Also Like” and “Traffic.”
Empty-Car Surfing
By Rollo Romig
Illustration by Erik Carter
Driverless cars will sometimes need to
move around the city with no humans in them at all — en route to pickups or drop-offs. As empty moving vehicles become a common sight, they’re sure to enable an annoying off-label use: a flowering youth subculture of external joy riding, similar to subway surfing.
All over the country, bored teenagers will lurk on the roadside, waiting to leap onto the roofs of empty cars, hanging on for as long as they can. The surfers will take advantage of advances in cloaking technology to help them confuse the cars’ sensors. Some will develop a clever assortment of grips and attachments to help them stay on, although the best among them will sneer at those who need to use tools. Fatalities will be inevitable, and often gruesome.
To keep the kids off, car owners will try to coat their cars in slick repellents; these will be effective but too messy to be practical. Soon, dealers of autonomous cars will offer optional rows of spikes similar to those we already use to keep homeless people off benches. Once again, design will become more hostile in the name of safety.
Rollo Romig is a journalist based in New York.
New Frontiers in Mobile Leisure
By James Somers
Just two months ago, the House unanimously passed a bill that would allow self-driving cars to be largely exempted from the safety standards that for decades have constrained the design of automotive interiors. There are only so many places you can bolt seatbelts to the car’s frame, after all. But once autonomous technology matures, some models will be made so much roomier that they will become rooms. A high clearance and thin frames will mean getting in without bending down. All the impedimenta of front seats — the jutting dash, the steering wheel and pedals, the high center console with a stick — can fold away. You’ll be able to swivel in your chair like a captain in “Star Trek.”
Liberated from having to watch the road, commuters’ eyes will be free to — what else? — consume video content. Sound’s hold on drive time — first with terrestrial radio, then satellite radio, then podcasts — will gave way to sight. But what kind of videos will we watch? If Marchetti’s constant holds (this is the idea that as transportation technology and urban planning change, people contort their lives to keep their commutes to an average of one hour per day), they will be not feature films but TV shows and short videos of the kind being increasingly produced for Facebook and YouTube. Podcasts will be displaced by vodcasts. Media companies are already negotiating with carmakers about how to price their offerings. Should in-car entertainment work more like cable or like the on-demand system in a hotel? Like Netflix or like Roku?
Most riders who play games will use their phones. But by the time one in five cars is autonomous, console makers will develop specific products for it. The nearly two hours a day of console gaming played by some 60 million Americans will tick up accordingly. Swarms of commuters hitting the morning and evening roads in waves will become swarms of druids or mercenaries in massively multiplayer online games, some of which might come to incorporate the driving experience itself, a Pokémon Go for the highway.
It seems less likely that we’ll use this opportunity to steal time back from screens. But if there’s enough space for a table, maybe commuters will sit around it playing poker or telling stories. Cars are, indeed, the most private places we have, sealed shut, out of earshot. (The writer Nicholson Baker has said that he likes to write while in a car parked in a shady spot.)
Chris Anderson — formerly the editor of Wired, now the head of a drone-data company and, in his spare time, of a racing league for autonomous cars — says the question of what we’ll do inside self-driving cars always comes up at cocktail parties. It always confounds him. “We have this world,” he says. “It’s called the back seat of an Uber.” We already know what passengers do in back seats, be they in a limo or an R.V. or a tour bus: They pull out their phones or their iPads, or they talk to one another, or they take a nap. Why should we think that not having a driver will change anything?
In an essay for Real Life magazine titled “Perpetual Motion Machines,” Chenoe Hart argues that we miss something when we think of autonomous cars this way, as a difference in degree, not in kind. Because we won’t have to worry about driving, vehicle interiors can expand to a point where we’ stop thinking of them as vehicles: They might have fully functional kitchens, or gyms, or beds. Instead of going to the coffee shop, the coffee shop might come to you; instead of going into the garage to pull out your car, the garage will be the car. The ability to summon spaces like this, Hart writes, will turn the built environment into a kind of random-access memory — with every destination equally accessible from anywhere. Whether it drove to you or you to it will be irrelevant; motion will be so smooth and constant as to be imperceptible. Hart argues that when travel becomes fully autonomous, we will always be traveling: families or co-workers in quiet concert on giant roads, moving in a cluster of moving buildings, chasing good weather, hardly aware that we’re going anywhere at all.
James Somers is a writer and programmer based in New York.
The Autonomous-Car Workout
By Gretchen Reynolds
Dashboard resistance straps would allow
mid-ride rowing.
A harness could
allow squats on a balance ball.
Ceiling straps would enable a classic
pull-down exercise.
Interiors could include exercise bikes
with harness clips.
This upper-body
chest fly would require straps anchored to the windows and doors.
Dashboard resistance straps would allow
mid-ride rowing.
Ceiling straps would enable a classic
pull-down exercise.
A harness could
allow squats on a balance ball.
Interiors could include exercise bikes
with harness clips.
This upper-body
chest fly would require straps anchored to the windows and doors.
There’s never enough time in the day to exercise, but perhaps the solution is in our cars. “A self-driving vehicle could definitely make a great gym, because it’s rigid all over,” says Marko Vujicic, an engineer and a partner at NPD Team, a company that consults with exercise-equipment manufacturers about machine designs. “That rigidity theoretically allows you to use every plane of the car against which to apply resistance. Your car becomes a full weight room on wheels.”
It’s not as straightforward as it sounds. For one thing, the vehicles would not be freed from the laws of physics. Should your car abruptly brake while you were, for example, running on a treadmill, the treadmill would cease moving forward — but you wouldn’t, at least until you hit a windshield or door.
It might be easier to remain seated. “The car could have special pockets all over the place that hold resistance bands of different strengths,” says Bryant Johnson, a personal trainer and the author of “The Ruth Bader Ginsburg Workout.” “You could use those to get a full-body strength-training workout.”
Throughout your commute, your future car could scrutinize your efforts, says Steve McCauley, the head of health services for the Wynn Las Vegas resort, with fingertip monitors or other sensors tracking your heart rate and the composition of your sweat.
But realistically, many of us will use our salvaged commuting time otherwise. In tomorrow’s cars, the area beneath the hood may not be needed for a combustion engine, Vujicic says — but it will
remain, for reasons of mechanical stability and safety. Fitted with some padding and a safety belt, it becomes a napping cave.
Gretchen Reynolds is the Well columnist for the magazine.
Limin Hee on Learning From Singapore
Interview By Jenna Wortham
If self-driving cars start to make up a significant portion of urban traffic, they’re likely to push cities to become more populous and compact, with fewer owned vehicles and more car sharing. The future, that is, would look more like Singapore, one of the world’s densest cities, whose government in recent years has tried to curb car ownership and ease congestion problems. We reached out to Limin Hee, the director of research at Singapore’s Center for Livable Cities, to ask her about how she thinks autonomous cars will begin to change urban life.
Tell me about Singapore’s approach to self-driving cars. The idea of autonomous vehicles is one measure we’re looking at to see a reduction in private car ownership. According to research done with M.I.T., we would only need about a third of the current vehicle population to provide mobility on demand. We find that a lot of private cars spend most of their time in parking lots. The land we use for parking could be used for other purposes. We are a very dense city, and we have been trying to plan our households around concentration of people. By 2030, we anticipate that eight out of 10 households will be within a 10-minute walk of a rapid-transit station.
What about people who don’t live in a dense urban environment? Eighty percent of our population lives in high-rise housing. So we can safely say the majority live in dense conditions.
How else do you see autonomous cars being used? Another possible use for them is trash collection and road sweeping. They don’t have to operate during people’s working hours. They can work in the dead of night, which helps us with congestion problems.
What are the other motivating factors? We also have to reduce our reliance on manpower. It’s been increasingly difficult to recruit truck drivers and bus drivers in Singapore from our local population. We’re also trying to reduce our reliance on foreign labor.
How will the technology transform the city? We are not so excited about the technology itself, as far as people owning self-driving cars. We are more focused on using autonomous vehicles as an urban infrastructure — a shared system that could allow the kind of accessibility that people have come to expect from their own cars. They could be as convenient as private cars, but at a much more affordable price point, and without adding to congestion and pollution.
How might the shift reshape the work-life balance? Our whole relationship to commuting will change. It’s very appealing to our millennial generation, who are constantly on their phones and screens, and it will free people from focusing on driving. They can email and go on social media or read a book.
You’ve talked a lot about the upsides. Are there any potential downsides that you’re preparing for? There are people who rely on being drivers as their jobs, but this is part of a greater technological disruption. There will be ways of either retraining or reskilling people whose jobs depend on driving to keep them gainfully employed. And if we have larger fleets of driverless cars, there will be more jobs to maintain these vehicles and keep them up to speed.
Will it be affordable? The price of public transportation is regulated and managed, but we do want to offer people a range of options, including trains and car-sharing, so people can pick the option at the price point that works for them.
In some cities in the United States, public transit has made segregation and social inequality worse: For example, certain train stops and bus lines don’t run from lower-income areas to wealthier ones. Is it possible that driverless cars will cause similar discrepancies? We actually think this will make Singapore more equitable as we try to move away from private cars, which worsen the divide between the haves and have-nots.
Jenna Wortham is a staff writer for the magazine.
The Blind Spot of A.I. Cars
By Yiren Lu
The ongoing transformation of cars — from mechanical systems of gears and switches into computers with dreams of self-determination — has been accompanied, perhaps predictably, by a parallel evolution in security threats. A decade ago, hacking a car relied on physically tapping into its internal network, for instance through a port sometimes found behind the radio. By 2011, it was possible to do the same thing remotely. In a series of demonstrations using internet-enabled navigation and entertainment systems as ways to hack into a Jeep Cherokee, Charlie Miller and Chris Valasek — both of whom later worked on self-driving security issues at companies like Uber, Didi Chuxing and Cruise — turned on its windshield wipers, disengaged the transmission and disabled the brakes, all from 10 miles away.
These exploits, though dramatic, took advantage of the same vulnerabilities in cellular networks that might plague any tablet or smart vacuum cleaner.
But a future in which one in five vehicles is autonomous will not only increase the attack “surface” (in the metaphor favored by researchers) of the car; it will also force manufacturers to confront more insidious security risks that are only now coming into view. It will become possible to compromise vehicles by hacking the environment.
A car that drives itself uses sensors to construct high-resolution images of its surroundings, then runs them through a “classifier” to identify all the objects in the image. The classifier can do this because it has trained on labeled images; after seeing a hundred thousand trucks, it learns to recognize certain features as indicating truck-ness. Today’s most popular and effective classifiers for object recognition, used widely in self-driving car software, are called deep neural networks.
But a deep neural network offers researchers very little understanding of why objects become classified as they do. Think of its output as a map, with each classification corresponding, like a ZIP code, to some demarcated region. Instead of existing in just two dimensions, though, the map has more like a thousand dimensions — and the slightest move in any one of these dimensions can push us into a different ZIP code and produce a false classification. This makes it possible for a hacker to construct slightly perturbed images, called “adversarial examples,” that look innocuous to the human eye but fool the car’s brain into making the wrong call, with potentially fatal consequences.
Adversarial examples are extremely difficult to defend against. “We don’t understand these attacks,” says Dawn Song, a professor of computer science at the University of California, Berkeley. “But the deeper question is that we don’t understand deep neural networks.” In a recent paper, Song and her co-authors introduced a method of constructing robust adversarial examples that can even be printed on paper and pasted to road signs. A stop sign defaced with specially calibrated graffiti, for instance, is recognized as a 45 m.p.h. speed-limit sign.
Song’s warning extends beyond road infrastructure. While researchers for decades have been working on — and developing countermeasures for — software vulnerabilities, the study of security issues related to deep learning is still in its infancy. “We don’t have a precise mathematical definition of a pedestrian,” Song says. “There’s no way for us to write down a formula describing exactly what a pedestrian is. So we can’t even really define our security goals.” Even with a recent surge of interest in adversarial examples, researchers are still struggling to understand the scope of the problem. Given the physical-world ramifications, Song says, “we should have started even sooner.”
Yiren Lu is a writer and software engineer in New York.
The Liability Conundrum
By Nathaniel Popper
Illustration by Tomi Um
If you think the aftermath of car crashes is complicated now — the arguments over fault, over repair costs, over whose insurance policy will pay — wait until self-driving cars are sharing the roads with human drivers. Start with all those delicate and expensive parts that communicate with a car’s software in order to make its autonomy possible. A fender-bender involving a Tesla means an expensive fender needs to be paid for; a part full of sensors and cameras costs a lot more to replace than a hunk of plastic on the back of an old Camry. In fact, some car insurers, like one AAA affiliate, have already said they would charge more to insure certain Tesla models. Tesla has countered with a campaign to convince insurers that its cars will be in fewer crashes; the company points to findings by regulators that its vehicles are involved in 40 percent fewer accidents when its autopilot software is installed.
The trickiest part of the transition to a self-driving future may be figuring out liability when accidents do happen. Instead of quarreling only about which driver was at fault, we will also be able to argue about whether the self-driving features in either car may have played a role. A preview of these complications was evident in the aftermath of the crash that killed Joshua Brown, the Ohio man whose autopiloted Tesla ran into a tractor-trailer. Federal investigators determined that the accident was not caused by a defect in the autopilot software, and Tesla has said repeatedly that drivers should always have their hands on the wheel even when they are using the technology. But officials noted that Tesla’s technology was unable to detect cross-traffic like that presented by the tractor-trailer. Tesla subsequently updated its software to make similar accidents less likely in the future.
So who was responsible for covering the damages? The insurance policy that covered Brown, who may not have fully understood the capabilities of his car, or Tesla? Neither Tesla nor Brown’s family has commented on who paid. And both car insurers and insurance regulators have been relatively slow to contend with the difficulty in assigning liability when a person is driving a car only part of the time. “Insurance policies have not necessarily caught up with the global advancement of A.I. technology in cars,” Brown’s lawyer told me.
Not surprisingly, Silicon Valley, which is racing to raise the curtain on the era of autonomous vehicles, is also thinking about insurance. A start-up called Driav intends to bring down insurance rates by monitoring the reams of data coming out of self-driving cars in order to figure out, on a second-to-second basis, who or what caused the errors that led to a crash. “It’s silly that claims are a ‘he said, she said’ argument,” Dan Peate, Driav’s chief executive, told me. “The data can just tell us what happened.” The company hopes to begin offering insurance policies next year, with the expectation that they will be cheaper in part because the cars will be in fewer accidents and in part because it plans to be able to shift some liability onto the carmakers.
In the long run, when all vehicles are driving themselves, the manufacturers themselves are very likely to assume all of the blame for accidents. Most such companies have said that they plan to pay for anything that happens while their cars and software are in charge. When that’s the case, insurance begins to look a lot more like a product warranty, covered by the carmaker, than a stand-alone product that a customer has to worry about. But in the transition years until then, there are likely to be a lot more messy arguments.
Nathaniel Popper is a technology reporter for The Times.
Get Ready for D.O.A.s
By Mark O’Connell
Illustration for Erik Carter
When I think about the future these days, I inevitably return to a tweet from the writer Elizabeth Hamilton-Argyropoulos (a.k.a. @bourgeoisalien): “The saddest part about self-driving cars will be all the times people die mid-trip and then ur dinner guests or pizza guy will arrive dead.” To read this joke is to become immediately certain that this will absolutely happen — of course it will! — and that it will furthermore become as much a part of the texture of everyday future life as, say, traffic delays caused by collisions are now. Admittedly, the dead-pizza-guy notion doesn’t seem very likely: A large part of the impetus behind the development of self-driving technology in the first place is the desire to do away with the need for flesh-and-blood people who demand wages for their work. But the image feels symbolically resonant precisely because of the ghostly future of all those moribund jobs. Self-driving cars bearing the earthly remains of the recently deceased — automated Ubers, say, with heart-attack victims sprawled incredulously across the rear seat — are, in this sense, an extreme metaphor for the near future of capitalism. The technology will be miraculous, yes; but if it functions as intended, it will serve as an uncanny vector of human obsolescence.
Mark O’Connell is a writer based in Dublin.
What Uber Was Built For
By John Hermann
“Ride-sharing” was always a misnomer for companies like Uber and Lyft. As with the rest of the “sharing economy,” nothing was ever being shared; labor and services were being bought and sold. But as ride-share companies have grown, their ambitions have multiplied and compounded. Currently, their horizons are limited by widespread car ownership,
alternative transit options and, primarily, their own cost. But autonomous cars make their grander plans — dreams of vast fleets of perfectly allocated and used vehicles — more plausible. They zero out a single variable, the cost of labor, that these companies believe is holding them back from utterly reshaping the world. Ride-share businesses are uniquely poised to manifest and demonstrate the big ideas of self-driving cars. They aren’t waiting for this change; they’re doing everything in their power to hasten it. Autonomous cars aren’t just part of the plan; they are the plan.
Conservative estimates put the world’s Ubers and Lyfts in an enviable position as self-driving cars roll out. Bullish projections put them in an unassailable one. A 2017 report by Tasha Keeney, an analyst with ARK Invest, told the story in four numbers. A rider’s “all-in cost per mile” for average United States taxi services was about $3.50. For a human-driven Uber in San Francisco, that number was about $2.86. Assuming the same fuel price, the figure for personal cars is much lower, at around 70 cents. The estimate for self-driving taxi services, circa 2020, would cut that cost in half.
At 35 cents per mile, tasks that were previously too expensive or that would have felt ridiculous become reasonable and routine. A world in which moving yourself or moving things to you costs almost nothing in terms of money or attention would remake places and transform lives. It would also bring with it new and complicated absurdities — the sort Americans adapt to quickly and then forget. (Like bottles of water shipped around the world and sold for $2, or electronics products assembled in one country from parts from six others, to be shipped to a seventh.)
Riders will thrill to this new infrastructure as it incorporates the roads around them, providing novel experiences and conveniences. But this excitement will prevent us from seeing that our aspiring fleet managers hope to create something like the profound consolidation of the internet’s recent past. Just as Amazon the online store laid the groundwork for Amazon the comprehensive and omnipresent service, apps for human-driven taxis may have laid the groundwork for a thrumming on-demand grid, on which people and things can be placed and retrieved.
Like Google and Facebook before them, which remapped, rerouted and dominated the old and diverse web infrastructure on which they were built, Uber and Lyft could say, credibly, that they work better with size and full participation. That market domination is merely the flip side of completeness. They will aspire to monopoly and could achieve, with startling speed, real and unprecedented claims over the roads we share — they will say they’ve found a better use for them, and if all goes to plan, riders will agree.
John Zimmer on A Whole New Lyft
Interview by Kevin Roose
Unless you’re a millionaire or a truck driver, your first encounter with a self-driving vehicle will probably be a car you ordered through a ride-hailing app like Lyft or Uber. These companies, which spent the first stage of their lives trying to get drivers into their cars, are spending the next one trying to get them out. There will be some initial awkwardness around self-driving vehicles, just as there was around ride-hailing in the first place. (Remember when it was weird to get into a stranger’s unmarked car?)
But Lyft, for one, is betting that it can put a trustable face on the tech. In July, it announced Level 5, a lab where it will work on its own driverless software, in addition to developing an open self-driving network that companies like Ford and Jaguar Land Rover will be able to plug their vehicles into. The idea of the network is to combine multiple kinds of autonomous vehicles under one routing-and-request system, so that when you order a Lyft, the app will assess your route, check the weather and
other factors and send you the best ride, whether that’s an autonomous Waymo or a human-driven S.U.V. John Zimmer, Lyft’s co-founder and president, explained its thinking.
Uber built its own self-driving technology. Why did you decide to do it the way that you did? We weren’t in a position to start this effort two or three years ago. But by being later to launch, we’re able to take a different approach. Many platforms and technology have changed a lot in the last couple years. There are things that we can skip over.
Lyft made people comfortable with riding in strangers’ vehicles by having pink mustaches on the front of the cars and friendly drivers. How do you plan to make people comfortable with a car that has no driver? I think that the first wave of autonomous vehicles will look very similar to the ones currently on the road, but as we move to the second wave, they’ll be more like a room on wheels. In a situation where you have 10 or 12 people in a car, we will have more of a shared-vehicle experience. In those cases, having a driver-host will make a ton of sense. Having hospitality delivered by other people is another way to make users comfortable — that’s why we don’t think there will ever be a moment where there would be fewer drivers or fewer job opportunities.
You don’t think that there might be fewer Lyft drivers even 20 years from now? I think you’ll have way more. The cost of owning a car is $9,000 a year. Let’s say we offer a $500 monthly plan in which you can tap a button and get access to transportation whenever you want it, and you get to choose your room-on-wheels experience. Maybe you want a cup of coffee on your way to work, or you want to watch the Warriors game later, so you’re in what’s basically a sports bar, with a bartender. If 0.5 percent of all miles driven are done on a ride-sharing app, and then if that number increases to, say, 80 percent, it’ll be such a huge industry shift that even if only 2 percent of that 80 percent is done by human drivers, it still represents a drastic increase in the number of human ride-sharing drivers.
Will there be a price difference between human-driven cars and autonomous cars? Maybe in the early days, people will pay more for the
service, and again in the long term, when it’s a full-on experience. There are a lot of variables.
When cars first replaced the horse and buggy, there was a backlash from people who thought they would take over. Do you think something similar will happen in the next few years as people start to see autonomous cars on the road? If companies don’t do a good job of explaining what’s happening and how it will impact society, that’s possible. I’m confident in how Lyft is going to behave.
Right now, you guys want to be everyone’s friends in this industry. Will there be a time when the business becomes more mercenary? We’ll see. We’ve never seen a shift like this in our lifetime. To think that only one or two companies will handle the transition is incorrect.
Kevin Roose is a business columnist for The Times and a writer at large for the magazine.
The Future of Car Sex
By Molly Young
For reasons that have gone unstudied, humans become horny in transit. With the proliferation of subway systems came the proliferation of subway gropers. With commercial air travel came the “Mile High Club.” Pamela Anderson and Tommy Lee filmed their famous sex tape on a boat (I’m told). There’s something about transportation that gets people in the mood. What is it?
My theory is simple. In our capacity as humans, we have three speeds at which we operate: crawl, walk and run. In other words: extremely slow, slow and slightly less slow. When we engage with equipment that increases our velocity, we become excited. This excitement applies in all cases, whether we’re talking about an escalator, a moving walkway, a bicycle, a roller coaster, a golf cart or a Chrysler LeBaron. It is fundamentally thrilling to travel at an unexpected speed. When the enabling mode of transportation is one that permits our excitement to be processed as erotic energy — even if that processing is specifically legislated against — some of us will do that. The distance between “excited” and “stimulated” is, after all
incredibly short. If there were a way to have sex on a bicycle, people would be bike-sexing all the time. It would have its own verb.
Having sex in a car is not currently impossible, but it is difficult. Hand-eye coordination, a mechanism critical for sex-having, is compromised for the driver, whose hands must remain on the wheel and eyes on the road. In road couplings, the driver is not an equal contributor. Unless, of course, the driver doesn’t need to drive at all. Which brings us to the self-driving car: basically a motel room on wheels. The one hindrance to this development is existing public-lewdness laws, which are fairly expansive. You can’t intentionally get naked and perform a “lewd act” in a public place without risk of penalty. Obviously. But you also run a risk if you perform your lewd act in a private place with the intention of being watched — say, in a self-driven automobile slowly cruising past a school bus where my future grandchildren are watching in horror as a frenzied orgy unfolds within smelling distance. I am confident, however, that the law will be updated to distinguish between acceptable and unacceptable car-based encounters.
And if not, industrial designers can come up with clever “solves,” the word that start-ups use for some reason instead of “solutions.” If car windows no longer need to function as navigating aids, why not tint them to inky obscurity? Or replace them with mirrors facing outward? Or mirrors facing inward? Or remove them entirely? Future riders may not even need a human partner in order to have an intimate appointment. Nissan is working with Microsoft to develop an “in-vehicle virtual personal assistant” which, if deviously hacked, could provide the 21st-century equivalent of phone sex.
More interesting than the aesthetics of car sex will be the cultural consequences. New slang terms will arise. New positions. New trophies. (How many times can you have sex in the time it takes to drive from Las Vegas to Los Angeles?) Virginities will be lost, children conceived, marriage vows consummated and violated. Car-based kinks will be minted and postcoital rituals established. New professional opportunities will emerge: interior decorators who specialize in “self-driving bedrooms”; entrepreneurs who launch apps to enable no-strings-attached in-car assignations. Preferences will be identified and explored. The world will expand in narrow but pulse-quickening ways — not a sexual revolution, but a plot twist.
Molly Young is a contributing writer for the magazine.
As a Partner and Co-Founder of Predictiv and PredictivAsia, Jon specializes in management performance and organizational effectiveness for both domestic and international clients. He is an editor and author whose works include Invisible Advantage: How Intangilbles are Driving Business Performance. Learn more...
0 comments:
Post a Comment