- Hello, from San Francisco.

This is Patrick McGee from the Financial Times

and I'm quite excited about this upcoming interview.

We've got 30 minutes with John Krafcik.

He is charismatic, he is energetic,

and he is the CEO of Waymo,

formerly known as the Google self-driving project,

which goes back to 2009.

They are really in, I mean, I think

pretty basically you could say

they're the leading self-driving / driverless

/ autonomous robot taxi service operating.

They operate the only real driverless service

in Phoenix right now.

They, I believe test in 25 cities, including on my streets

it's even possible that I'll flip the camera at some stage

if a Waymo goes by.

They've done more miles autonomously

than any other driverless company.

And if we really want to get a sense of

when this technology might change our lives,

I think it was nobody better to speak with

than John Krafcik CEO of Waymo.

Who is joining me here.

So hello, John.

- Hey Patrick, it's great to be here.

Thanks so much for having us.

- You got the Waymo mug very nicely placed just over your-

- Yeah, Waymo mug is strategically placed

and over my other shoulder,

there's the little Firefly prototype.

- Oh, wait, I can't see it.

I feel like I see a globe.

Oh, I see just above that, yes.

All right, we'll get to the Firefly in a second

'cause that's really interesting.

Yeah, this is your no pedals, no steering wheel vehicle

but you've moved beyond that.

And actually that's something that I wanna get to

in a little bit.

I'm hoping you can begin

with what I consider to be a pivotal story for Waymo.

Which is you developed your first driverless car

in I think late 2009, 2010.

And by 2012 you were doing these pilot projects,

allowing Google employees,

So people that had nothing to do

with the self-driving project,

you're allowing them to commute with their vehicles.

Sorry, with your vehicles.

Telling them, you know under no circumstances,

should you not be overseeing the vehicle.

You need to be doing that, but look how cool this is.

You know, let's do some tests and you filmed them.

And I think the film, when the footage came back to you,

it shocked you and it really

put Waymo on a different trajectory.

So do you wanna tell us that story?

- Yeah, yeah and maybe just the tiniest bit of context too.

So you're right, Patrick.

The project started in 2009 and by 2010,

this scrappy little team of 20 to 30 folks

had done some pretty extraordinary demonstrations

of fully autonomous driving.

Famously there were 10, 100-mile challenges

that the team was able to compete

in the first year and a half or two years.

One of them included driving all around your neighborhood

in San Francisco and then over to Lombard Street.

There was another challenge.

They went from Mountain View up El Camino, all the way

222 stoplights or whatever it was to San Francisco.

The team drove around Tahoe, the team drive to Santa Cruz.

So they made a lot of progress

in demonstrating the promise of the technology, right?

And so at the time Google's thought was,

well, what's the first viable at scale commercial product?

The initial hypothesis was ironically

something called autopilot, team called it autopilot.

And it was meant to command the car very safely

to drive the car very safely

from entry point to a highway or freeway to exit point.

And so in Google fashion at the time,

the company asked for volunteers within Google,

we call this internally in the Google world, dogfooding.

And lots of people were interested in this possibility.

You'd get a free car at the time, the team was putting

the driving mechanisms and the sensors

and the compute onto Lexus RX SUVs.

And there was a great demand for folks

to actually give this a try.

So the bar was set quite high.

You had to agree to certain stipulations

including an indication that you understood

this was beta technology and it might not be perfect.

So you needed to stay at vigilant.

You could take your hands off the wheel

but you had to keep your eyes on the road.

You had to stay alert

and we'd have cameras in the car to monitor you.

And if we saw you misbehaving,

we would take away this great privilege, right?

So we started the project sometime in February,

I think it was 2013.

And within a month we shut the project down

because we saw so many examples of humans misbehaving.

And it's this fundamental conundrum that we face

whenever humans are forced to supervise technology, right?

It's really hard

and as the technology gets better and better

and by the way this technology

that we had at that time was amazing.

Humans tend to check out and just assume that

the technology is going to be perfect.

So in our videos, which you can find in,

Patrick, if you haven't seen them we can send them to you.

I think they're somewhere on our website at waymo.com.

We saw some first indications of concern with

the folks in the driver's seat turning around

to fuss with things in the backseat.

We saw one woman putting on makeup, using an eyelash curler.

And the scariest thing we saw was a Googler

who was driving to work early in the morning, predawn,

driving down highway 280 at about 62 miles an hour

who fell asleep because they had so much confidence

in this technology that had been working for them so well

over the course of a week or so, right?

They had already checked out.

So we shut that effort down and it sort of inspired us

to move in a different direction to solve for full autonomy,

fully autonomous driving.

- Okay, so let me pause you there.

The reason I'm bringing up this seven-year-old story

is I feel like it has added urgency now

because Tesla, which I guess to some extent is a rival,

has its own technology also called Autopilot.

But the latest iteration is called full self-driving.

And I feel like it's pretty clear they've come to the same,

you know, fork in the road.

They've seen the same problems with automation, complacency.

And yet they've basically said, you know,

"We're not taking liability, the driver is."

They have some prompts that they agree to

and we're gonna allow the system to run.

And they want to expand it, you know eventually to

potentially hundreds of thousands of vehicles.

So based on Waymo's decision,

I wanna know to what extent you think that's reckless.

Because I don't know that regulators and consumers

really make distinctions between

different self-driving systems.

And what worries me is that

if mistakes get made on Tesla's part

and you know, these are cars traveling 60 miles an hour

so clearly fatalities could be involved.

Does that not pose a risk that you cast upon

on the entire self-driving industry?

- Yeah, maybe a couple points to make there, Patrick.

It is important that we talk about these things.

The first is Waymo's mission in the world,

isn't to be a car company.

Our product is a driver.

That's our sole focus.

And if you look at the business lines

that we are just now starting,

for example the ride sharing service that is fully open.

Unlimited availability to anyone who's in Southeast Phoenix,

you can hail a Waymo.

Just download the app and a fully self-driving Waymo

will come in and take you from wherever you wanna go,

from wherever you are to wherever you wanna go.

So the technology is here right now.

Our key technology is the driver.

That's the most important point.

That's what we're here for.

We're not a car company, therefore,

so we really don't see Tesla as a competitor.

Rather we see Tesla and other car companies

working primarily in the driver assist area,

which is important and good.

And good driver assist technology can save lives.

There's no question about it.

The challenge, I think for the auto industry,

the traditional auto industry, is to ensure that

consumers understand the limitations, right?

And the conundrum that we saw at Google back in 2013 is,

as the driver assist systems get better and better

and better, humans will tend to have more propensity

to check out and not do as good a job as

supervising that technology.

It's a challenging conundrum.

So the good news right now is that the driver assist systems

do need human attention and they require

constant surveillance and humans are able to stay

sufficiently busy for the most part,

monitoring those things.

As they continue to get better, though that's the challenge.

You would think that there would be increased safety

but there's also increased risk at the same time

that the human licensed driver in the driver's seat

might check out at just the wrong moment

when the car needs some help.

- But do you worry about Tesla being reckless

and posing risks that might come back to haunt

the likes of Waymo?

- I think, you know, it's nothing that we can really control

at the Waymo side.

We're gonna do our best to speak about our technology

and deploy it safely and responsibly.

You know, I do think it's important

for all the participants, both on the driver assist side

and the fully autonomous side

to be as precise as possible with language, right?

And if a licensed driver is required,

it should be referred to as a driver assist system.

If a licensed driver in the car isn't required,

which is the only technology that Waymo is working on,

then I think then you should call that

a fully autonomous solution.

- Right.

Okay, so given that you went over some of the early facts

of having a autonomous driver system

that could navigate Lombard street,

you know, famous tourist, windy hairpin turn street

in San Francisco 10 years ago.

What in a sense is taking so long

for this to sort of conquer cities the way that Uber did?

I think a hundred cities within four years.

And I guess what I look back to with Waymo is

in 2018 you ordered 60,000

or up to 60,000 Chrysler Pacificas

and up to 20,000 Jaguar, I‑PACEs.

But if I look for the latest statistics,

the latest I've heard is still around 600 vehicles.

I think you probably just haven't updated it

and it's more than that,

but correct me if I'm wrong, it's not in the thousands.

It's not in the tens of thousands.

So why the postponement in delaying a driverless solution?

- Look, I think the technical challenge

that we're talking about is probably the most complex thing

that a group of humans have ever tried to do.

Moving a large physical mass from any point A to any point B

on the ground with all of the chaos and entropy

that's associated with traffic

is an extraordinary task, right?