In 1956, GM unveiled a high-tech concept car it called the Firebird II. If ever there was a Car of the Future, this was it. The sleek four-seater was powered by a jet engine, featured a titanium body, and sported a central tail fin that would have done the Batmobile proud. But the car’s most futuristic feature wasn’t visible on the outside: The Firebird II could drive itself, or at least, GM promised, such a car would be able to navigate unassisted in the not-so-distant future.

GM released a slick promotional film showing how an all-American family would take a road trip in the Firebird II in the halcyon era two decades hence: After a brief radio conversation with the “control tower,” Dad steers the car into the “high-speed safety lane” and is instructed to take his hands off the wheel. “You’re now under automatic control,” the tower informs him. Dad stows away the steering wheel and lights up a cigar while his wife and two kids enjoy cold drinks from the onboard dispenser. Their jet-powered car serenely steers itself down an almost empty highway, passing desert landscapes and elegant buildings straight out of The Jetsons.

Of course, it’s easy to make fun of old, overly optimistic predictions about the future. But over the past decade, enthusiasm for self-driving vehicles has come back with a vengeance. Only this time the concept is at least technologically plausible. And it is backed by serious money. Some of the world’s richest companies—including Google, Tesla, Uber, and Apple, not to mention Ford, Toyota, and, yes, GM—have sunk billions into the autonomous-vehicle (AV) race. Today, some 60 companies are road testing AVs and associated technology in California alone.

So should we all expect our cars to start chauffeuring us around within the next few years? Not so fast. Despite all that research and investment, the dawn of the AV era keeps getting pushed into the misty techno-future.

There are many factors that drive technological change. One is the state of engineering: What is technologically feasible? But there’s another, often under-appreciated factor: What is desirable? How much change do consumers really want? In the 1950s, futurists predicted that in 50 years few people would choose to cook; they would just thaw out precooked meals. And no one would want clothes made of wool or cotton; synthetics were so much more practical! Those futurists would have been shocked to see 21st-century hipsters knitting sweaters, or modern dads still cooking meat over hot coals like their Paleolithic ancestors did. Convenience and efficiency aren’t everything, it turns out.

So it is with self-driving cars. The push for autonomous vehicles isn’t coming from actual car owners. It’s coming from carmakers, green-transportation advocates, and, especially, Big Tech. In fact, the vast majority of drivers don’t want to have anything to do with robot cars. According to a 2021 survey by AAA, 86 percent of respondents said they would be “afraid to ride in an automated vehicle.” So if car buyers don’t want self-driving cars, why are automotive and tech firms pouring so much money into the field? Oh, they have their reasons.

For automakers it’s about money of course. Right now, their revenues mostly come from selling us new cars. But since a typical car now lasts about 12 years, manufacturers don’t get too many bites at that apple. Carmakers would rather get paid for the time you spend in a vehicle, much the way Uber makes money on every ride you take. That’s why Ford is testing AVs in partnership with Lyft, Toyota has partnered with Uber, and so on.

These companies envision a future in which fewer people drive their own cars. Instead, they see a world in which we opt to ride in fleets of autonomous, all-electric cars and vans that are almost constantly in motion, shuttling paying passengers from place to place. Green-transportation advocates also like the idea of replacing owner-driven cars with shared vehicles, which would effectively turn the automobile into a decentralized form of public transportation.

For Big Tech, it’s all about the data. “If the car is doing the driving, then you’re free to spend more time online, shopping or working,” former Car and Driver editor Eddie Alterman told me. “They want to colonize more and more of your time so they can harvest that data.” Ultimately, that data might be more valuable than the vehicle you travel in. Instead of being a bastion of privacy and independence, the car then becomes yet another digital panopticon we willingly inhabit. “The key metaphor for the car has always been liberty, the freedom to go wherever you want,” Alterman says. “But that’s becoming less and less true.”

In many ways, we are already giving up some of our individual agency when we drive. New cars increasingly come with digital tech meant to make driving easier and safer—lane-departure warnings, smart cruise control, collision avoidance, and the like. The most advanced of these systems, such as Tesla’s so-called Autopilot, can almost entirely take over the task of driving in predictable environments such as major highways. Drivers must keep one hand on the wheel and are told to stay alert to hazards. But, not surprisingly, studies show most drivers quickly lose focus when Autopilot is engaged.

AV boosters say such driver-aid systems will soon evolve into genuine full autonomy. Currently, when semiautonomous systems like Tesla’s encounter confusing traffic patterns or bad visibility, they are programmed to shut down and hand control of the vehicle back to the (hopefully wide-awake) driver. Higher levels of automation—what the Society of Automotive Engineers calls Levels 4 and 5—would mean the vehicle could drive itself under most or all conditions. (“Level 5 means you can go to sleep in the back,” one expert says.) A Level 4 AV known as Cruise, a joint venture between Honda and GM, is currently being tested in San Francisco. (The test rules do require that a “safety operator” ride shotgun, ready to flip the kill switch if needed.)

But experts I’ve talked with think the transition to full autonomy is less of an incremental progression than a leap into the unknown. “It’s not primarily a technical problem,” Alterman says. “It’s a psychological problem.” Today, we accept that human drivers are fallible. While we don’t excuse accidents caused by driver error, our society has come to tolerate them as a cost of our car-oriented lifestyle.

We see machines differently. Riding in a fully autonomous car means putting our faith in hardware and algorithms we don’t fully understand. That requires a much higher level of trust. “Riding in an AV is more like getting on an airplane,” Alterman says. “You know that you are not in control.” But at least in an airplane, you know the pilots care deeply about keeping you—and themselves—alive. A robot driver, on the other hand, doesn’t have self-awareness or a sense of self-preservation. It’s not morally concerned if something goes wrong. That means a fatal accident caused by a bad algorithm or a shoddy sensor strikes us as somehow scarier than one caused by a drowsy or reckless driver. In theory, if every car on American roads were autonomous, highway accidents would plummet. But until we reach that supposed Shangri-La, every death caused by AV technology will loom especially large in the public’s awareness. “Until we have had zero AV fatalities for a very long time, people aren’t going to trust this technology,” Alterman says

And it’s not as if AV companies have ironed out the technical bugs. Teslas operating on Autopilot crash with alarming regularity. Real-world AV tests have been plagued with accidents. Most are minor fender benders. (Since AVs tend to slam on the brakes at the first hint of a problem, they often get rear-ended by human drivers.) But in 2018, a pedestrian was struck and killed by an autonomous Uber vehicle undergoing testing in Phoenix. (The “safety driver” didn’t even look up from her phone.) Perhaps frustrated by slow progress and legal headaches, Uber last year sold off its AV division at a “fire-sale” price. And Uber isn’t the only company to learn that teaching computers to drive cars is much harder than they expected.

The biggest challenge AV systems face is predicting the behavior of all those pesky humans clogging up towns and cities—other drivers, cyclists, jaywalkers. No AV firm has yet demonstrated a fail-safe ability to navigate chaotic city streets. Until humans and robots learn to get along better, this challenge could delay the full roll-out of AVs for years, perhaps indefinitely.

“The bottom line is, unless you have all autonomous vehicles, you really can’t have any autonomous vehicles,” Alterman says. For some AV advocates, that’s an argument for banning human-driven cars from some urban centers, or for setting up autonomous-only lanes on highways (like that “safety lane” in the 1956 GM film). In this view, there’s nothing wrong with algorithmically controlled vehicles. It’s those unpredictable humans who make all the trouble. “There is always a danger that tech companies will convince cities to re-make their streets in terms of infrastructure that benefits AVs and not pedestrians,” Nicole Gelinas, of the Manhattan Institute, warns. More likely, we will see AVs first used as people movers in closed environments, such as college or corporate campuses. Perhaps AV vans will cruise protected routes in some cities, augmenting bus or light-rail services. And AV trucks might operate in a handful of dedicated highway lanes in a few years. Those sorts of narrow applications could have economic and environmental benefits.

But Americans, in all their anarchic, freedom-loving glory, are unlikely to submit to having too many streets or highways reserved for our robot overlords. A world prioritized for AVs would be one in which we’ve lost the independence and spontaneity of driving a car—not to mention walking or biking—on our own streets. “Your personal autonomy will have been sacrificed to the machines,” Alterman says. Technology can do a lot to make our world more convenient and safer. But we—as individuals—have to remain in charge.

We want to hear your thoughts about this article. Click here to send a letter to the editor.