In a suburb of Phoenix, Arizona, there's a fleet of 600 minivans shuttling
people from place to place.
Ordering one feels almost exactly like calling an Uber, except for one thing:
the vans are driving themselves.
It does feel like from when I first got in that it is just a normal car.
Their vehicles are all over our roads.
I don't think you could stand at a street corner or drive a couple of
miles without seeing a Waymo vehicle.
I know that some of this technology is scary for many of our citizens, but
I think if you see other times in our economic arc that this has really
opened up new worlds for people and new opportunities.
This is my thirty third year in policing, and when I started in policing we
had vehicles, obviously, but there were there were no computers, no cell
phones. Pagers weren't even in existence yet.
So to see this technology in relatively a short 30 year span is just
In 2004, the U.S.
Department of Defense hosted a 142 mile driverless car obstacle course
The farthest any of the entrants got was a little over seven miles.
And off we go.
Good try, guys.
The next year, the DOD tried again, this time five vehicles completed the
course, but a team from Stanford did it the fastest.
That Stanford team was led by a computer scientist named Sebastian Thrun.
So I didn't anticipate this to become a race for speed. It was one
of the most thrilling races ever.
In 2007, Google hired Thrun and he created Google X.
Two years later, Google X launched a self-driving car project.
In 2016, that project spun off as its own company, called Waymo, under
Google's parent company, Alphabet.
Here's an example of the Waymo Chrysler Pacifica minivan.
There's nobody driving.
There's nobody behind the wheel.
The company says it's tested its vehicles in over 25 cities in six states.
But the most miles seem to have been driven around Phoenix, Arizona.
From the police department's point of view, our mission is to keep our city
safe. So we recognize this technology as something that could really
impact our roadways because the overwhelming majority of collisions are
You get in the car and you have a seat and it has a start button.
And it's pretty trippy when you can see the fact that the car is driving
It's amazing to see how well the brain processes information, as a driver,
to see the car do the same exact thing.
It's great to be a part of history, for my kids to experience.
My daughter actually liked it a lot, didn't she?
Right now Waymo is doing two main things in the Phoenix area: Around a
thousand members of the community have access to its rideshare beta
program, Waymo One. Users can summon one of Waymo's 600 vehicles 24/7 and
ride anywhere within a limited local region.
And these users are actually paying for the rides, it's not just a free
demo. It also has a partnership with Lyft and makes ten of its vehicles
available to the general public via Lyft's app.
I probably use Waymo maybe percent of the time.
The biggest limiting factor is that it only goes in a certain defined
area, mostly in Chandler and Tempe and maybe a little bit of Mesa.
If it went all the way downtown, I would probably take it a whole lot
The reason Waymo is limited to a small region is because its cars are
autonomous, but only in specific locations.
Everywhere it can drive has been carefully mapped and analyzed so that
even before sensing anything new, the vehicle already has a good sense of
where it is.
The Society of Automotive Engineers came up with a set of standards
defining the levels of autonomy a vehicle could be, ranging from zero to
five. And right now, Waymo's vehicles are at a four: capable of full
autonomy, but only sometimes.
Tesla refers to its driver-assist systems as Autopilot.
Nobody in the industry thinks that's the case.
Waymo and General Motors Cruise Automation are very close to having what
they refer to as level 5 cars, most of the time.
So right now it's standard for Waymo vehicles to have safety drivers behind
the wheel at all times, ready to take over if something were to go wrong.
And beyond that, there's a team of support staff on call to help riders.
The vehicles are constantly maintained by a team of people.
They're cleaned by a team of people.
While the driving itself is done mostly by a computer, the system is still
dependent on human labor.
A lot of the business promise and also the hope for these machines, these
autonomous vehicles, is that they eliminate labor and they eliminate the
need for human beings to drive and to be stuck in jobs like delivering
pizzas or picking up the elderly or the blind from their homes and taking
them to services, wherever it needs to be.
When we think through that a little more carefully, though, some of the
chinks in that idea show up. For
example, think of something like Meals on Wheels.
The vehicle shows up.
It opens the door.
There's the meal.
Maybe they can get out to the curb to get it, maybe not.
Even if they could, though, when that human driver shows up with a Meal on
Wheels, they actually come to the door.
Maybe they sit for a little bit.
So it's this human interaction that's still very much a part of these
I think there's still a ways to go before they're ready for prime time on
But we want to be helpful in the testing of it.
And then we want to make certain that, whether it's at the state level or
the federal level, that all of those regulations and rules are being
Developing vehicles that adhere to strict safety protocols, including speed
limits, has occasionally been a point of contention for other human
drivers on the road.
There's been some experience where because our Waymo vehicles actually
follow the rules and the law, that some people who tend be in a rush, get
bothered by that.
So there's a transition that's going to happen.
But with rideshare companies like Lyft and Uber struggling to be
profitable, for them, leaning into self-driving cars could make sense.
As we saw from the Lyft and Uber IPOs, there does not appear to be a path
to profitability for ride hailing services with human drivers.
Even buses where, you know, operating the vehicle is very expensive, the
vehicle itself, the major portion of that expense is the driver.
The future of autonomous vehicles is more likely to be in the form of
ridesharing fleets that you can borrow when you need, but no actual car
ownership. So I think they see an opportunity in cars that will be able to
transport things, transport people, but not so much around car ownership.
And it's still a little bit unclear as to where they see the biggest money
coming from, but at least that's where it's evolved to.
In March 2018, a woman named Elaine Herzberg was killed by an Uber
self-driving car just 13 miles from Waymo's office.
But it didn't slow Waymo down.
It was business as usual in Chandler.
The next day, just as many vehicles were on the road.
It was an unfortunate incident for another company.
But again, Waymo has had an extremely conservative business model and
safety protocols that they had really weathered that storm well.
We were very saddened, of course, by what happened in Arizona.
Our hearts go out to the family and all those impacted by the crash.
At Waymo, our focus has always been safety.
In our city, there have been no collisions where the Waymo has been at
fault. So you can take that any way you want as an indicator, but it's
such a small sample size.
Certainly we anticipate the more these vehicles are out there functioning
at the level that they're expected to function at, if it takes away that
human element, it potentially could have a very positive impact on the
Some people have asked, you know, is it actually safe?
You know, when you are inside, do you get nervous or, you know, do you
think anything is going to go wrong?
And I'd say, well, you know, no, there's always a driver, you know, at
least while they're still getting the technology, you know, hammered out.
Waymo is way ahead of everybody else in terms of the technology.
They have these disengagement reports in California.
They disengage a lot less, a lot fewer times, than anybody else.
You know, I think the robot drivers are probably actually better than human
Arriving shortly at your destination.
Please keep your seatbelts fastened until we reach your destination and
remember to take all your belongings with you.
Proponents of autonomous vehicles make compelling claims about the
potential benefits of self-driving cars.
94% of all crashes are due to human error.
42 hours are wasted sitting in traffic per person per year in the US.
That's an entire working week every year.
And millions more people aren't able to drive because they're elderly or
living with a disability.
And self-driving cars have the potential to change all of that for all of
I think these cars and automobiles and trucks provide a real opportunity
for the state.
There's a lot that can be done for disabled people, for blind people, for
So many of the deaths that happen on our roads are a result of human
error and I believe these autonomous vehicles can provide higher public
safety and that really is the objective.
But it's just not clear these things would actually happen with more
self-driving cars on the road.
One of the things I often hear from people is when an autonomous vehicle is
better than the fiftieth percentile driver on the road, we have an
absolute responsibility to let them onto the road.
Others, like Elon Musk, have said it's almost irresponsible not to have
these vehicles out there because they are safer and will be safer than
That's not been proven.
It presents a problem, which is people dying on the road or crashing and
so forth, and saying, well, therefore, you need this solution.
But of course, there are a lot of solutions.
And one of the solutions we see right now are things like autonomous
braking, lane keeping assist, all of these driver-assist systems which
take a good driver and make them better.
And so even if we could say that an autonomous vehicle was better than a
human driver, it doesn't mean that an autonomous vehicle is better than a
human driver plus all the advanced driver assist systems we have.
And if the goal is safety above all else, there are other less complicated
things that could be done.
For example, since speeding is known to be one of the top causes of car
accidents, members of the European Parliament recently provisionally
agreed to require all vehicles sold in Europe to include mandatory speed
A lot of the promises about autonomous vehicles are around congestion and
They are kind of a silver bullet, Silicon Valley, a tech-bro solution to
the problem of road deaths.
There's a much less exciting solution to road deaths, particularly in
urban areas, and it's called Vision Zero.
And the premise is pretty straightforward.
It says, let's start with safety and then let's add mobility.
The current idea around driving, around cars, is let's get as much
mobility as we can and then let's start to make things safer.
Whether or not autonomous vehicles are safer than human drivers is in a
lot of ways beside the point.
They're more lucrative than selling cars to people.
They're more lucrative than selling rides driven by human beings.
So while there are other, potentially better solutions, updating
infrastructure and making policy changes is never going to be as
interesting to most people as cars that can drive themselves.
With the new technology, there's going to be a time period where you have
to, you know, give it a try and work out the bugs.
Like if there's a computer program, I don't think I've ever seen somebody
code something and hit run and it works perfectly the first time.
You have to give us some real world experience.
And so, you know that not everything is going to work perfectly right off
I used to say a year ago that I was sitting in a diner and looking out the
door at 6 a.m.
and I saw in the span of an hour 12 Waymo vehicles.
That was trumped about six weeks ago when I saw about 30 Waymo vehicles at
intersections. And I don't know if it was a parade or whatever, but it
was, they're just all over our streets.
And it's a good relationship.