Ready or Not, Driverless Cars Are Here (2023)

This transcript was created using speech recognition software. While it has been reviewed by human transcribers, it may contain errors. Please review the episode audio before quoting from this transcript and email transcripts@nytimes.com with any questions.

From “The New York Times,” I’m Michael Barbaro. This is “The Daily.”

Today, after a closely watched vote, driverless cars, once a Silicon Valley fantasy, have become a 24-hour a day reality in San Francisco. My colleague, Cade Metz, describes the unique challenges of co-existing with cars that drive themselves.

It’s Wednesday, August 23.

Cade, we have invited you in our studio here in New York to tell us about what’s happening in San Francisco. Which doesn’t sound very efficient, but be that as it may, San Francisco has become the capital of this grand experiment of driverless cars. And I want you to just describe what it’s like to live in San Francisco in that new reality.

As you walk through the streets, there are hundreds of these cars driving around almost constantly in most of the city. Once you get out of the downtown area, the Financial District, you cannot walk a block without seeing one of these cars. And if you’re a first-timer in San Francisco, or you’ve arrived for the first time in years, you cannot believe what you’re seeing — hundreds of cars driving around with no one in the driver’s seat.

These cars are not hard to spot. They’re outfitted with all sorts of sensors designed to detect everything that’s going on in the world around them. These are big, sometimes trashcan-sized sensors that spin. And you can’t —

I’m getting an image of some hybrid between a police car and a kid with braces.

An analogy I like to make is the Ghostbusters car, the car from the original “Ghostbusters” 1984, which has all this stuff all over it.

OK, so you’ve been in these cars. I want you to just describe the whole experience of getting in the car and how you even get one of these driverless cars. Because obviously it’s not just a regular old hail system.

Let’s go to my most recent drive, which was last Wednesday. I pulled open my Waymo app, where I can hail a car, much as you would an Uber. The difference was, it wouldn’t pick me up exactly where I wanted it to pick me up. It gave me a spot in an alleyway about a block and a half away, where there’s less traffic.

And eventually my car pulls up. I pull the app out again, and there’s a little button on the app that says unlock.

As I step into a car, a disembodied voice comes over the speaker that welcomes me by name. “Hello, Cade.”

It also gives me a short safety briefing.

Then the wheel starts to turn on its own. The car starts to move. It’s like being in the car with my daughter just after she had turned 16 and was learning to drive.

On edge!

Yes, and no. I’m on edge because my daughter is a new driver, but also I have a daughter who is wonderfully conscientious. She is so careful when she drives she would stop several feet before she got to the intersection. But that’s what these driverless cars are like. They drive at the pace and with a care of a particularly conscientious 16-year-old driver.

[LAUGHS]:

As the drive continues up and down the hills of San Francisco, it is taking much longer than it would have taken in an Uber because of that overly cautious nature of the car.

[MUSICAL TONE]

So overall, how do you rate this ride, this ghostly, driverless, cautious version of a taxi ride?

It got me to where I wanted to go, but it took an awful long time, relatively speaking, to get me there. Now, the flipside, it’s also like a ride at Disney World, where it’s a real novelty. It’s something that you don’t necessarily experience every day. And you take this ride with your senses heightened, aware of this strange thing that you’re doing.

And like a lot of novelties injected into the real world like this, it is the subject of extreme debate, particularly in San Francisco, where these cars have been causing accidents. In some cases, those accidents cause injuries. They run into emergency vehicles. They do things that they are not supposed to do.

And increasingly, that is a concern for many of the residents of San Francisco who don’t want their city to be a Petri dish for these types of cars that may cause serious problems with daily life.

Right. I’m thinking back to what you just said. It’s like a ride at Disney World, but San Francisco is not Disney World. It’s a city with public streets and pedestrians. And so that makes me want to understand how we have gotten to this moment where an experiment on the scale that you are describing is even possible. So tell us the story of how we arrive at this moment, where San Francisco is filled with these ghostly driverless vehicles.

Scientists and researchers have been trying to build autonomous vehicles since the ‘60s. The aim, and it is largely driven by labs funded by the Defense Department, is to build autonomous vehicles for the military. It’s better to have a vehicle, a tank, that does not have a driver behind the wheel, so to speak.

In case it gets shot at, destroyed.

Exactly. That’s the rough goal. But as time goes on, people start to realize that you can do this for civilians as well, that you can use the same technology, and use it to drive an ordinary car, and potentially change our lives as well, potentially build a car that is safer than a human being.

Technology does not get tired. Technology does not drink whiskey and get behind the wheel. There are all sorts of advantages to having technology drive as opposed to humans if you can get technology to the level where it can drive as well as we require.

And when does that start to happen?

It starts to happen around 2005. DARPA, which is the research arm of the Defense Department, organizes a contest. And the idea is to get all sorts of research teams to build a driverless vehicle, and then compete with each other to see whose is most proficient. And two guys notice that this is happening.

They happen to be the founders of Google, Larry Page and Sergey Brin. One of the leaders of one of the research teams, a guy named Sebastian Thrun, tells the story of Larry Page showing up at one of these competitions in disguise — a hat, sunglasses, so he won’t be recognized.

What ends up happening is that Page hires Thrun to run a new lab at Google called Google X. And the idea is to go after what Page and his colleagues call moonshots, extremely ambitious technical projects. The first thing they decide to go after is a driverless car.

The ultimate moonshot.

The ultimate moonshot. Larry Page wants to put one of these in your driveway, to put you in a car on a daily basis that can drive by itself, get you to where you want to go while you’re doing something else.

Taking a nap, reading to your kid.

Absolutely. And that’s something that excites a lot of people. But there’s an interesting difference of opinion between Thrun and Page that explains the next 10 years of development with this technology. Larry Page, running a company like Google, wants to see this happen.

Sebastian Thrun, the man he’s hired to run this lab and to run this project, he realizes how difficult this will be. He realizes how far they will have to go. And he doesn’t necessarily know how to get there.

And why is it such a challenge? Why in Thrun’s mind is it going to be so hard to realize Larry Page’s vision of a driverless car in every driveway?

Getting a car to make a left turn at an intersection is not difficult. A right turn is not difficult. Accelerating, braking, those things are not difficult. What’s difficult is dealing with all the chaos that we encounter every time we get into a car the unexpected is likely to happen at any given second.

You have to have a machine that can completely, first of all, recognize what is going on around it. You need to know that that is a pedestrian and not a bicycle. Then, once you are clear on everything that is around you, you need to predict what’s going to happen with each of those things. And then once you predict what’s going to happen, you need to develop a way of responding to all that. And that requires reason. It requires a quality that machines still don’t have, and certainly didn’t have in the early 2000s.

Right. Driving a car isn’t just about steering the wheel as you go down the road. It’s about responding to all the possible things that could go wrong. Dog runs out into the road, bike swerves into your path — we as humans learn to respond instinctively to those. It’s harder to teach a robot to do that.

That’s right.

But despite how challenging teaching a robot all that is, we know that this driverless car technology does move forward. So tell us how that happened.

Larry Page, the founder of Google, says go after this. And Thrun goes after it.

Right. Jump, how high?

So in 2009, Thrun and his colleagues at the new Google X lab start building a driverless car. And by 2010, they quietly start testing this technology on the public roads of California.

A secret driverless car project that puts those driverless cars on the road?

Until “The New York Times” and my predecessor here John Markoff got wind of the project and in the fall of 2010 let the world know that it was happening.

And what is the response?

Well, it sets off a tech arms race, not only among dedicated technology companies like Google, but among a new type of company which is doing ride hailing, Uber and Lyft, and the traditional car companies who now see a threat to their way of doing things. They jump in and start competing as well.

And as companies like Google and Uber started to talk about this technology publicly —

They talked about it as if it was around the corner.

If you were reading the tech press, if you were reading the mainstream press around that time —

— you assumed that by the following year —

You would have a driverless car in your driveway if you wanted one.

And that was never the case.

There was some hype.

There’s always hype in Silicon Valley. In this case, the hype was particularly strong.

OK. So now that all these big players are pumping this extraordinary amount of money into this unproven technology, what becomes of it?

On the surface, the types of cars being built were perfect. Reporters would get into a car in Mountain View, California. It would take them around the block with nary a mistake. But the reality below the surface was that building this technology was extremely difficult, and that continued for years, where the hype did not reflect the reality. But then that changed in 2018.

Which is when?

There was a crash in Arizona involving an Uber test car. A 49-year-old woman at night was crossing a multi-lane road with her bicycle. She was struck and killed. There was a safety driver behind the wheel who was supposed to take over if something went wrong, but they did not.

The bottom line is, someone was killed by a self-driving car, and that almost immediately set the industry back. Yes, people behind the wheel are involved in accidents every day. People die because of those accidents. But if a new technology kills someone, that is far more concerning to the general public, to regulators. They start to wonder if this is going to work.

After the crash, Uber immediately pauses its efforts to test these vehicles, not only in Arizona, but in California and in Pittsburgh. Toyota follows suit. And slowly, all these companies start to admit that this is harder than they had let on.

Even Google starts to change the way it’s doing things. Its self-driving car operation is spun out into a new company called Waymo. And its ambitions start to morph away from that idea of a car in your driveway towards the type of robotaxi service that Uber and Lyft were building. Rather than having a car in your driveway that can potentially drive anywhere, what these companies decide to do is, let’s limit a self-driving car service to a particular geographic area.

Narrow the variables.

That’s the way it works. If you put a ring around San Francisco, you limit your cars to that one city, you can build a digital map, a literal 3D visual map of the city, and you can give that map to the car. And then you test over and over and over again in that constrained environment.

You limit the chaos, you limit the unexpected moments. That’s going to be easier, and that’s what the industry starts to go after. And that’s how we get to the point where San Francisco, the second most dense city in America, becomes a testing ground for these driverless vehicles with hundreds of cars driving around the city.

This goes on for years, where multiple companies are testing these vehicles to ensure that they are safe. But as they test —

— it becomes more and more clear to the people riding the cars and to other citizens in the city that they are flawed —

— that they do make mistakes —

— that they can gum up traffic —

— that they can cause accidents —

— that they can cause injuries.

All of this comes to a head with a public hearing earlier this month in California, where the future of these companies in this city will be decided. The question on the table is, can they operate in the city like an Uber or a Lyft, only without drivers behind the wheel?

But this decision is even bigger than that. If these companies don’t get approval in San Francisco, the flagship of their operations, what happens in other cities across the country?

In a sense, this is a meeting where the future of the driverless car is at stake.

Absolutely.

We’ll be right back.

So, Cade, set the scene for us at this meeting in San Francisco at which nothing short of the future of the driverless car may be on the line.

This is a meeting where four state regulators are going to decide the future of two different companies, Cruise, which is owned by General Motors, and Waymo, which is owned by Google. And San Francisco is a particularly interesting place for this type of vote. Activism is very much a part of this city, and has been for decades, and you’re going to get people who do not hesitate to voice their opinions, their concerns over this technology.

But at the same time, this is the center of technological innovation in the country and maybe the world. And you’re going to get people who are equally adamant that this should happen and should happen now.

Those two sides of San Francisco come together on August the 10th at this public hearing.

Where regulators are going to decide the future of this.

You get a group who are all for Waymo and Cruise starting paid rides immediately.

This ranges from bicyclers —

— to people with disabilities, medical professionals.

And all sorts of other reasons for why these cars should be allowed on the roads immediately.

It sounds like above all, they’re making an argument around safety.

Safety is the prime argument, but there are other arguments.

My name is Sean Durkin. I am a rider in driverless vehicles. As a gender nonconforming person, I have experienced rideshare drivers who have left me on the side of the street, refused to open their vehicles.

That this technology does not discriminate against people in the way that a human driver would, a taxi driver might pass someone up on the street.

And then there are people who stood up and just gave the classic Silicon Valley justification.

Technological progress has to happen.

Your poodle be damned.

Exactly.

Thank you, commissioners. My name is Michael Smith, and I’m the technical co-founder of a successful startup here in San Francisco. And I strongly support tech when it’s useful, safe, and actually works.

Then, of course, there are just as many arguments on the other side.

People who argue that this is not as safe as human drivers.

That it’s going to prevent fire trucks and police cars from getting to where they need to go.

People also argued, as they often argue as new technology is rising up —

This will take away existing jobs.

That Uber drivers who now earn their living driving around the city will be replaced by driverless cars.

And you get all sorts of people showing up —

— who are just angry at the technology industry.

That anger has been building for many years in San Francisco —

— as tech workers take over the city, really change the fabric of the city.

And the cost of the city.

And the cost of the city. And some people are just fed up, and this is an outlet for them to voice their anger.

Right. Because what could better embody the kind of disembodied power of tech and its money and all that perhaps people don’t like about that than literally a driverless car?

Sometimes it’s hard to pin down technology and the changes it is making. This is a physical technology in a physical place that is causing real chaos in the city.

OK. So all of these people testify, but ultimately, as you said, just four people are actually going to vote. So how does that vote actually go?

In a vote of 3 to 1, the commission votes in favor of Waymo and Cruise to allow paid driverless rides across San Francisco immediately. And this is a sign that these companies can also move ahead across the country with their ambitions.

And in that sense, they’re opening the driverless Pandora’s box. So what happens after this vote?

We see inside this Pandora’s box almost immediately. Within a week, there are two accidents involving Cruise that really open people’s eyes. One of these Cruise cars does not properly recognize a construction zone and drives into wet cement. Talk about a moment that can be easily shared over and over again on social media.

And I assume it was.

It was indeed.

Now, there have been incidents like this for years in the city involving these cars. But amidst this vote, there’s added scrutiny to the technology, and people start to notice what is going on.

Right, or what’s going wrong. Right.

A few days after the cement incident, there’s an accident involving a fire truck in San Francisco, where a passenger in a Cruise vehicle is injured.

The very next day, the California DMV asked Cruise to cut its driverless car fleet in half.

Wow.

So they drop it from about 400 cars down to 200.

Well, that’s interesting. So the state commission votes to allow this technology to expand, and then the DMV turns around, what, a couple of days, a week or so later and says pare it back. What do you make of that?

Well, it shows what has been obvious for a long time is that these cars are flawed, and it’s all about how much tolerance we as a society have for those flaws. In this path towards this utopia where technology can do everything on its own and make the world safe, how much chaos, how many accidents, how many injuries are OK as we reach for that goal?

Well, I guess that’s a good question. How much tolerance should there be when it comes to driverless cars? Because the incidents you just mentioned, the car going into the wet cement, surely that’s embarrassing, but does it fundamentally change our understanding of the safety of these cars?

And isn’t this kind of glitchiness the natural course of technological advancement? There’s a period of transition. It’s rough until eventually there’s widespread adoption. Or is the case of the driverless car special? Is there something about taking a human being out of a two-ton moving piece of metal that means our tolerance is just going to be lower? How do you think about that? How should we think about it?

Our tolerance is absolutely going to be lower. That’s the way our minds are wired. We as human beings trust human skill and ingenuity, and we trust our own judgment to make the right decision when lives are at stake. Do we trust this technology, which in so many ways is not as adept as we are, that cannot reason like we can, does not have the common sense that you and I really depend on and depend on others to have?

It does not have that. It has many other things that we do not. It has more sensors. It can process data faster in some ways, but it doesn’t have that thing that makes us human. Do we trust that thing that is so different from us to replace us? That is a hard question for anyone to answer. And that’s what we’re trying to answer here.

Right. And I think that raises the question that goes all the way back to the beginning of our conversation and to that moonshot that Google’s co-founder Larry Page introduced. The question being, is this a moonshot, the driverless car that any of us really need?

That is still an open question. We don’t know whether we really need this. And even if we are sure of that, it’s unclear whether the technology can satisfy that need. We don’t know yet if it’s safer than a human driver. We don’t yet know if it’s cheaper than a human driver, which it also needs to be.

What’s going on in San Francisco is an opportunity to answer that question and to see whether or not the need is there and whether or not the technology can satisfy that need.

And when are we going to the answer?

[LAUGHS]:

I won’t hold you to it.

It took 13 years to get to this point. It’s going to take years to get to that point where we can really answer that question.

Well, Cade, thank you very much.

Thank you.

We’ll be right back.

Here’s what else you need to know today. “The Times” reports that Ukraine’s counteroffensive against Russia may be faltering because it has positioned too many of its troops in the wrong places. The central goal of the counteroffensive is to cut off Russian supply lines in Ukraine’s south, but US military leaders say that Ukrainian commanders had divided troops and firepower between Ukraine’s south and east, effectively spreading them too thin.

And eight candidates have qualified for tonight’s first Republican presidential debate in Milwaukee, which will not feature the race’s frontrunner, Donald Trump, who has refused to participate. The candidates include Governor Ron DeSantis of Florida, Trump’s former vice president, Mike Pence, former governors Chris Christie and Nikki Haley, Senator Tim Scott of South Carolina, and the conservative businessman Vivek Ramaswamy. The debate will begin at 9:00 PM Eastern.

Today’s episode was produced by Rikki Novetsky, Olivia Natt, and Luke Vander Ploeg, with help from Shannon Lin, and Jessica Cheung. It was edited by Devon Taylor with help from Michael Benoist, Paige Cowett, and Lisa Chow, contains original music by Marion Lozano, Rowan Niemisto, and Dan Powell, and was engineered by Chris Wood. Our theme music is by Jim Brandenburg and Ben Landsverk of Wunderly.

That’s it for “The Daily.” I’m Michael Barbaro. See you tomorrow.