In 1896, describing the advantages of the cars he manufactured, R. E. Olds—inventor of the Oldsmobile—drew a contrast that Americans understood: “It never kicks or bites, nor tires on long runs. It doesn’t require care in the stable and only eats when it’s on the road.” Thomas Edison praised cars in similar terms: “The money spent in the keep of the horses will be saved, and the danger to life will be much reduced.” By 1930, cars had largely replaced the horse-drawn carriage as the dominant means of transportation in the United States, particularly in cities.
A century of progress in auto technology has made car travel easier and more comfortable in countless ways, but the basic economics for a hired ride hasn’t changed much. The cost of a typical taxi today is about the same as it was in 1870 to hire a horse-drawn cab, adjusted for inflation. Why? We eliminated the care and feeding of the horse, but the care and feeding of the driver has picked up the slack.
The time has come to finish the job by eliminating the driver.
Robotaxis—self-driving vehicles for hire—are about to eat the drivers’ lunch. Within a few decades, hiring a driver to operate a car will go the way of human elevator operators, bowling pin setters, and switchboard operators. In those cases, people were needed to perceive the relevant aspects of the situation and perform simple physical tasks in response. Advances in artificial intelligence, along with painstaking analysis of edge cases, has finally made it practical to do the same for operating a car. And the effects on our economy, health, environment, and lifestyle are likely to rival the impact of the automobile itself.
It’s been a long, slow slog for companies seeking to enter this market, with many casualties. Most notably, General Motors abandoned their Cruise robotaxi service after running afoul of regulators, while other car companies, such as Ford, have refocused on driver assistance. The barriers to entry include the enormous up-front cost of development and testing, the need to design specialized vehicles, and regulatory delays.
But there’s light at the end of this tunnel. Today, residents of several U.S. cities can summon a driverless car as easily as they can order an Uber or Lyft—sometimes even through these companies’ apps.
Currently, Google’s Waymo One service is the leading company in this market in the United States. Waymo provides more than 250,000 paid robotaxi trips each week in Phoenix, San Francisco, Los Angeles, Atlanta, and Austin, and plans to expand to ten additional cities this year.
Tesla is gearing up to give them a run for their money. The company recently started service in Austin, using current models equipped with “Full Self Driving” software, and expects to expand the fleet with its two-seater “cybercab”—which lacks driver controls—once it can get regulatory approval. Amazon-backed Zoox is also looking to start commercial service in Las Vegas this year.
You may wonder, of course, whether allowing computers to drive cars around our streets is a prudent idea. But if you live in any of the places where they are up and running, you already know the answer: an unequivocable “yes.” Robotaxis zipping around these cities make occasional mistakes, as human drivers do, but rarely cause accidents.
A recent study by insurance giant Swiss Re backs this up with cold, hard numbers. Analyzing liability claims for property damage and personal injury over 25 million fully autonomous miles driven by Waymo, they found an 88% reduction in property damage claims and 92% reduction in bodily injury claims compared to human drivers in similar circumstances.
Waymo vehicles in the study were involved in just nine “contact event” property damage claims and two bodily injury claims, while human drivers would be expected to have 78 property damage and 26 bodily injury claims over the same distance. Tesla safety data from the first quarter of 2025 shows similar results.
But even this isn’t nearly the whole story: The figures include accidents where Waymo was not at fault (claims, not awards). Waymos, like other cars, are often rear-ended while stopped or are otherwise passively involved. To date, Waymo has not been found at fault for either of the two bodily injury claims—a near perfect record. If only robotaxis roamed the streets of San Francisco, it’s plausible that car accidents might be virtually eliminated.
Pedestrians are beginning to notice. Villi Iltchev, founder and managing partner of VC firm Category Ventures, has learned to trust the technology. He was quoted in Forbes as saying, “As a cyclist, when I ride my bike and I get next to a Waymo, I know it watches me, and if I try to pass it on the right, it makes room for me. I feel so much safer because it always sees me. It will never get in my way. It will never cut me off. It will always prioritize my safety over itself.” A recent study by J.D. Powers showed that confidence in robotaxis soared from roughly one third of their survey sample to two thirds after taking a ride.
Then there’s the economics. How much of your money is going to the taxi or Uber driver? It’s a difficult question to answer because the driver is often responsible for paying a portion of the costs. But most studies suggest that the driver takes home roughly 50% of the fare after expenses. And that’s only the beginning. Fully autonomous taxis don’t need space for the driver and controls, and don’t have to haul around the associated extra weight. Ark, an investment research and management firm, estimates that the cost of operating a purpose-built robotaxi will eventually fall to around 25 cents a mile, compared to about 80 cents today for a personal car and $2 for a human-driven taxi.
Baidu, the robotaxi market leader in China, already charges only 55 cents per mile for its “Apollo Go” service in ten cities, where it reportedly operates more than a thousand robotaxis.
As robotaxis expand to more cities, suburbs, and rural areas, many consumers will conclude that it makes no more sense to own their own cars than it did for people to keep their own horses after 1930. Most private vehicles are in use only 5% of the time, while robotaxis can run around the clock without the need for driver breaks, naps, or meals, potentially increasing vehicle utilization by an order of magnitude. The annual cost of owning and operating a car in the United States today is over $12,000, and it’s typically driven about 13,500 miles. If robotaxi service eventually costs say 50 cents a mile, this pencils out to a saving for the average two-car family of more than $10,000 a year.
That said, there’s likely to be some reluctance. For many families, cars double as mobile lockers, playpens, dining and media rooms, not to mention trash containers. But the same technology driving robotaxis is also coming for your cherished ride. Why rent a precious parking spot in your apartment building when you can summon your own car in seconds from a nearby parking depot that occupies underutilized real estate—possibly the same facilities that serve as robotaxi staging areas? Own a typical house with a two-car garage? It takes up around 600 square feet, room for two additional bedrooms with baths.
But wait, there’s more good economic news. If the Waymo statistics carry over to personal self-driving vehicles, the cumulative effect could potentially slash your insurance payments and increase your productivity—since you can work while commuting. (Tesla already offers insurance discounts based on how much you engage their self-driving features.) What’s more, the real payoff of driverless technology is in reduced property damage, medical care, and lost wages. Traffic accidents cost U.S. businesses and consumers nearly $410 billion a year. Then there’s the small matter of alleviating the incalculable human pain and suffering of 2.1 million injury accidents and 40,000 excess deaths.
Savings of this magnitude are life changing both for individuals and society. What will people do with this annual bonus? Spend it, of course, lifting our overall standard of living. Don’t cry for the surplused drivers—this bounty will increase demand for goods and services, not to mention create new jobs such as cleaning, servicing, and “rescuing” malfunctioning vehicles, as well as managing staging facilities.
Then there’s what this will mean for our urban spaces, roads, public infrastructure, and environment.
About 14% of the incorporated land in Los Angeles is devoted to parking—nearly 200 square miles. If increased vehicle utilization and remote storage reclaimed half of that, the value of the repurposed real estate could exceed $140 billion in that city alone.
Twentieth century automobiles didn’t come into their own until horse-driven carriages and pedestrians were banned from roadways and traffic controls were introduced, allowing them to travel at far higher speeds. The rise of “motor culture” made it practical to live farther from work, creating the suburbs and highways while transforming our recreational, tourism, and vacation patterns. Banning human-driven cars from some or most roadways may have an equally dramatic impact.
Robotaxis (and other self-driving vehicles) offer a way to greatly increase the transportation capacity of our roads while decreasing travel time, since these vehicles can be spaced closely together at high speed, potentially eliminating our $127 billion in annual public spending on expanding highways alone. Tired of waiting at red lights? Future automated vehicles will be able to communicate amongst themselves and our public infrastructure, dynamically negotiating who has right of way and warning each other of impending hazards.
Does this mean the end of mass transit? Probably not. Utilization might increase substantially if local governments offer discounted or free robotaxi shuttle service to trains and other transportation hubs. In 2024, Waymo offered a $3 credit for rides to eligible Bay Area transit stations.
Fewer cars on the road and the fact that virtually all robotaxis are going to be electric means reduced energy consumption and pollution. Electric cars use about half the energy of gas-powered cars, even when generated by fossil fuels. Charging facilities can be centralized, diminishing the need to install them in residences and businesses.
Safety and savings aren’t the only factors likely to drive the transition to robotaxis.
In San Francisco, it can take less time to summon a Waymo than getting your own car out of the garage. But that’s not the main reason Sarah Price, an interior designer in Pacific Heights, finally gave up her car in favor of Waymo. After all, she can call a taxi, Uber, or Lyft. “I value my privacy. I’d rather not share my ride with a stranger, listen to their music, smell their takeout, or worry that they’ve been up driving all night. Not to mention that I’m too polite to explain that I’m not there for a chat,” she said. “For me, taking a Waymo is a few precious moments of peace and serenity.”
For similar reasons, some San Francisco parents have turned to robotaxis to take their kids to and from school, despite rules to the contrary. One parent was quoted in the San Francisco Standard as saying, “They wouldn’t feel comfortable in an Uber or Lyft, but they feel perfectly comfortable in a Waymo. I feel really safe because I can track it. It’s just awesome. I love it.”
So how is all this likely to play out?
Over the next several years, robotaxis will become commonplace in cities around the United States, along with self-driving personal cars. Once consumers get a taste of the economic and practical benefits, they will spread to the suburbs. Leading the advance into rural areas, small-business entrepreneurs will likely purchase and operate their own fleets, until it becomes cost-effective for the majors to either buy them out or take over their markets, as happened with dial-up internet access prior to broadband.
Car-owning families—a remarkable 92% of U.S. households—will simply refrain from replacing their aging second cars, then, as appropriate, do the same for their primary one. What about those weekend skiing or camping trips? Suitable self-driving vehicles will be available for rent by the hour, day, or week. Just order one up, throw in your gear, and enjoy your break. Need a pickup or van to move some furniture? Same story, but you can click an option to include some hourly workers to help load it up. Looking to save a few bucks on your commute? You can reserve a semi-private seat on a specialized multi-person vehicle, likely on a subscription basis. Your regular latte will be ready and warm when it arrives.
How long this transition from car ownership to “transportation as a service” (TaaS) will take is difficult to predict, but as a starting point, today’s average car is in service for 12 years. So I’d speculate that half the miles driven by private (and commercial) vehicles in the United States will be automated within 20 years. And as the related infrastructure is built out and exclusive “autonomous driving zones” expand their footprint, within another 20 years driving your own car on public roads will be considered as rude—if not immoral or illegal—as smoking in an elevator. (“Thank you for not driving!”)
By then, “private car” will mean riding solo, and a “driving range” will be where people go for weekend fun to try their skill behind the wheel, similar to horse stables today. And I doubt that tomorrow’s kids will mourn the dreaded teenage ritual of taking the driving test.
Robotaxis herald a sparkling new age of transportation, but it’s hardly the only facet. In conjunction with the coming wave of electric short-range passenger aircraft—which are quiet enough to operate in urban areas and can take off and land vertically—the modern transportation revolution will be complete.
It’s time to stop horsing around and kick the driver to the curb.
Jerry Kaplan teaches social and economic impact of Artificial Intelligence at Stanford University.
Follow Persuasion on Twitter, LinkedIn, and YouTube to keep up with our latest articles, podcasts, and events, as well as updates from excellent writers across our network.
And, to receive pieces like this in your inbox and support our work, subscribe below:
 | A guest post by
|