If you haven't seen a driverless autonomous vehicle circling your city or cerita lucah rakan sekolahneighborhood block yet, that may change in the very near future.
Waymo, the industry-leading robotaxi company, currently operates in Phoenix, San Francisco, and Los Angeles, and is coming to Atlanta and Austin via an Uber partnership later this year. Waymo has also made exploratory "road trips" to more than 25 other cities in a quest to become what it calls "the world's most trusted driver."
Other AV companies, like Zoox and May Mobility, are on the road, too.
This may leave many wondering whether a transportation revolution is underway, especially as consumers' personal vehicles gain more autonomous features, like Tesla’s self-driving — but supervision-required — vehicles. But the reality is far more complex, with tension between the technology's promise and how it's playing out in the real world.
When will I own my own autonomous vehicle?
First, let's get one thing straight: There is no car that you can buy from a dealership today that's fully self-driving, says Jeff Farrah, CEO of Autonomous Vehicle Industry Association (AVIA).
Yes, that includes Teslas. While the company's CEO Elon Musk recently said he planned to releasean unsupervised version of the car's full self-driving software later this year, drivers currently need to closely monitor their vehicle when engaged in FSD supervised mode.
What people can purchase today are vehicles equipped with lower-level autonomous features, like lane-keeping assist, adaptive cruise control, and basic navigation.
"I think it's important to distinguish that, at those levels of autonomy, there's still a massively important responsibility for the human driver to be able to take over at a moment's notice," Farrah says. (AVIA's members include AV operators Waymo and Zoox, as well as traditional auto manufacturers like Ford, Volkswagen, and Volvo. Tesla is not a member.)
The results when drivers don't take over can be fatal. In April 2024, a Tesla owner put his car in what the company bills as its FSD (Supervised) mode, then proceeded to look at his phone, according to NPR. Within minutes, the Tesla driver had struck and killed a motorcyclist, despite a warning from Tesla's system to pay attention.
Dr. Missy Cummings, a professor of robotics and engineering at George Mason University who reviewed investigation documents from that crash for NPR, told Mashable that she doesn't believe individual ownership of AVs will happen in her lifetime.
"There's a really good reason for that," Cummings says. "Even if we do finally solve some of these [safety] problems, you actually have to take such good care of these cars."
The leading self-driving cars use numerous sensors, including LIDAR and RADAR, to create a picture of their surroundings, among other critical tasks. These sensors have to be kept clean and calibrated for peak performance and safety.
So if you can barely keep up with your car's maintenance now, imagine trying to take care of an AV.
SEE ALSO: Mercedes-Benz beats Tesla to selling Level 3 autonomous cars in the U.S.If I can't own an autonomous vehicle, when will I be able to ride in a robotaxi?
Given the deployment of AVs in major cities, Kathy Winter, May Mobility's chief operating officer, says we've reached an "inflection point" in the last year. With significant investments from Toyota, the company is testing its driverless Sienna minivans in Ann Arbor, Michigan and Sun City, Arizona. It offers paid rides to the public in Peachtree Corners, Georgia, a city in the Atlanta metro area.
Amazon-owned Zoox just began testing its robotaxi in San Francisco. For now, only company employees can catch a ride in the vehicle, which looks a bit like a toaster on wheels.
Franklin Trujillo, who works on commercialization as the head of technical program management for product infrastructure at Waymo, agrees that there's been a rapid shift in robotaxi availability in the last year. The company has 150,000 paid rides per week, according to Trujillo.
Read more about riding in a Waymo: Smooth, silent, strange: What it's really like to hail a robotaxi
"This is real,"he says. "This is a fully operating service ingrained in daily life."
Waymo currently only provides rides to paid passengers on surface streets, but it recently began sending driverless cars onto L.A. freeways with company employees on hand to observe and share their feedback.
It took a decade for robotaxis to get to this point. While the progress is promising, you're still not going to suddenly see the AVs across the United States. That's because getting them on the road is far more difficult than the average potential rider probably realizes.
How do AV companies decide where to deploy?
Some places are simply harder to drive in than others, thanks to road and weather conditions, and potential limitations of a company's AV technology. Zoox's leadership, for example, said last yearthat they wouldn't be rolling out in snow anytime soon.
Even in San Francisco, where the weather is relatively mild, rain can challenge Waymo's sensors, says Jeffrey Tumlin. He regularly hails Waymo rides and recently served as the director of transportation of the San Francisco Municipal Transportation Agency. The role gave Tumlin a front row seat to all the highs and lows of living — and managing transportation — in a city with a robotaxi fleet.
Tumlin notes that San Francisco, with its complex geography and transportation system, posed problems for Waymo that didn't materialize to the same extent in Phoenix, where the streets are straightforward, there are fewer pedestrians and cyclists on the road, and fewer anomalous events, like a motorcade or protest.
So if an AV company isn't in your city yet, it may have to do with the nature of where you live. But Trujillo says that as Waymo learns distinct city layouts and characteristics, it can better manage similar challenges in the next place it arrives. Going from San Francisco to Los Angeles, for instance, wasn't that difficult, Trujillo says. He also notes that Waymo focuses on entering the top ride-hailing cities, where demand for a robotaxi would be strong.
What doesn't change from city to city is the detailed work that AV companies say they put into partnering with locals, which can include transportation officials, emergency responders, policymakers, and community groups.
Then there's the red tape. Some cities and states don't have regulations that permit AVs to operate on their streets. Waymo works with policymakers to change that. May Mobility has partnered directly with departments of transportation or other local agencies and businesses to both test their robotaxis as well as offer on-demand ride-hailing, which can be its own lengthy process.
Polls show that peopleare skeptical of AVs prior to ridingin them. In order to build consumer confidence in their product, some of the AV companies have been very deliberate about where and how they deploy, and emphasize the importance of safety.
"Unfortunately for all of us, there have been a few accidents that have set the industry back in terms of trust," says Winter, May Mobility's COO.
Are AVs safe?
Perhaps the most notorious accident involving a self-driving AV happened in San Francisco, in October 2023. A Cruise vehicle struck a pedestrian, dragging her 20 feet before stopping. The victim survived, but Cruise suspended its operations soon after and was ultimately fined $1.5 millionby federal regulators. Majority owner General Motors decided in December 2024 to shutter Cruise as a robotaxi service and instead focus on developing the technologyfor fully autonomous personal vehicles.
True believers in AV technology tend to see accidents like Cruise's as a rare exception to an unproven high-stakes rule: A driverless car powered by artificial intelligence and sophisticated sensors is unequivocally safer than a human driver.
But experts who study AV safety say it's far too early to make such pronouncements. Cummings conducted a recent study of safety datareported by AV companies between December 2021 and November 2023 to the state of California and found a number of concerning issues related to "unexpected driving behaviors." This can include collisions with stationary objects; the AV inexplicably disengaging, which stalls the car; and phantom braking, when an AV suddenly slows or stops without explanation, but most likely because it hallucinated an obstacle. Phantom breaking particularly worries Cummings, as she sees no signs yet that AV companies have solved that problem.
Safety incidents sometimes show up as a viral news story about a robotaxi that acted strangely or dangerously by heading into oncoming traffic, driving in loops around an airport, or rolling into wet concrete, but many more occur without becoming a headline. Additionally, AVs appear to struggle when they encounter emergency responders and traffic control, says Cummings. Some have blocked the path of ambulances and firetrucks, and others have gotten stuck in traffic because they can't interpret hand signals by a police officer directing traffic. (Waymo says its cars can interpret and follow hand signals and handheld signs.)
Waymo and Zoox are both under investigation by the federal government for unexpected driving behavior. Last May, the Department of Transportation's National Highway Traffic Safety Administration (NHTSA) opened an investigation into Waymo for 22 reported incidents in which the company's AVs collided with objects like gates, chains, parked vehicles, and also appeared to disobey traffic safety control devices.
Dr. Missy Cummings, who previously worked as the senior safety advisor for the National Highway Traffic Safety Administration, saw near disastrous accidents caused by human coding mistakes.
No other details are available because the company appealed to NHTSA to redact its response, citing confidential business information. Trujillo said he could not comment on the investigation, but he touts Waymo's recordof fewer airbag deployments and reduced injury-causing crashes and police-reported crashes compared to human drivers going the same distance in a city where the company operates.
"The data is clear, we are safer," Trujillo says.
Yet Cummings argues that it's very difficult to compare AV performance to human driving behavior for a number of reasons. Human drivers tally 3 trillion miles on U.S. roads each year, whereas AVIA reports that its members have driven more than 70 million fully driverless miles to date. There's also the matter of where they're driving: Highways see more deadly crashes than surface streets, and robotaxis currently operate exclusively in slower speed zones.
Cummings also points out that you cannot take human error out of coding algorithms that power AVs. Cummings, who previously worked as the senior safety advisor for NHTSA, saw near disastrous accidents caused by human coding mistakes, a lesser-known but important aspect of potential AV safety issues.
Related Stories
- Meet the autonomous driving influencers of TikTok
- Protesters decry 'fascist Elon Musk' at dozens of Tesla showrooms nationwide
- CES 2025: Here's John Deere's new driverless fleet
- Jimmy Kimmel pranking his aunt with a driverless car is a fun watch
- Waymo’s driverless cars are coming to these cities next
The RAND Corporation has estimatedthat driverless cars will need to tally 275 million miles in order to confidently demonstrate their failure rate, a benchmark they're far from reaching despite significant progress.
David Kidd, senior research scientist at the independent nonprofit Insurance Institute for Highway Safety, says that while Waymo's transparency is laudable, more research needs to be done so that the field can reach a consensus about what qualifies as safe when comparing human drivers and AVs.
"Can we make the claim that they're safer than human drivers? I don't think that there's enough evidence yet to put that stake in the ground," Kidd says.
Cummings credits Waymo for doing a "pretty good job" making their rides relatively safe so far, but worries that market pressures will produce a race to the bottom as shareholders and investors look to make a profit. Alphabet, for example, has spent billions funding Waymo.
"That kind of financial pressure, as a person who does AI and safety, that's a perfect recipe for an accident," Cummings says.
Elon Musk recently announcedthat Tesla plans to debut its own robotaxi service in Austin this June. Cummings said she worries people will die inside and outside of any Tesla robotaxi deployed so soon, given the company's safety record. Musk previously called Cummings "biased"against Tesla, and his followers subsequently harassed her aggressively online.
One significant but unanswered question about the future of robotaxis is how they'll be regulated by the Trump administration. With Musk overseeing efforts to gut federal agencies, safety advocates are worriedthat NHTSA might dismiss the directive that requires Tesla and other AV-operating companies to report crashes to the government. DOT issued this directive in 2021 in response to safety concerns, and Musk and other Tesla executiveshave been critical of it.
Regardless of what happens to that directive, Michael Brooks, executive director of the Center for Auto Safety, says that industry groups like AVIA have floated suggestions for regulatory standards. He hopes industry groups bring more consumer advocates, academics, and independent safety researchers into the standard-setting process.
"We're not really guaranteed anything coming out of that process other than what the industry's version of what safe is," Brooks says.
The future of robotaxis
When Lyft and Uber debuted more than a decade ago, they promised a lot, including reduced congestion and reaching underserved communities and riders with disabilities. Tumlin says those and other promises haven't materialized.
In fact, data showthat Lyft and Uber haveworsened transportation conditionsin some cities. Instead of getting people out of their cars, research shows the ride-share services have lured people off public transportation, like buses and trains. He worries the same thing will happen with robotaxi services, especially as they come under increasing pressure to demonstrate their market viability.
"I'm not seeing ways in which AV technology solves any of the problems that Uber and Lyft faced in the marketplace," Tumlin says.
"There's a lot of potential upside, but at the current stage of development of autonomous vehicle technology, we are not seeing any net benefit to the transportation system as a whole," Tumlin says. "We're seeing a good deal of userbenefit."
In other words, robotaxis in San Francisco are terrific for individual riders, but not transformative for the city.
He adds that the riskiest drivers may be the least likely to take a Waymo, because the behind-the-wheel chaos is a feature not a bug for them. So even if robotaxis are exactly as safe as the average human driver, the service could still increase overall traffic and thereby increase traffic risk as a result.
SEE ALSO: Uber gives users option to contact ADT with safety concernsSo in one version of the future, robotaxis replace human drivers at a massive scale while operating in diverse rural, suburban, and urban communities, serving everyone, including those with disabilities, while also dramatically increasing safety. But perhaps just as likely is a future where we have yet another service that does little to solve our congestion and safety problems but caters to well-heeled city residents who prefer not to talk to a ride-share driver, or who feel safer without one.
It's hard to know which future we're going to get, or if a hybrid of the two will emerge. It's easier instead to look out the window of a robotaxi and assume that a billion-dollar technology will get us where we reallyneed to go.