Aicha Evans, Chief Executive Officer, Zoox Interviewer: Emma Hinchliffe, Fortune
Category
đ€
TechTranscript
00:00Hi.
00:00Good morning, Aisha.
00:01Thank you so much for joining us.
00:02Good morning.
00:03Thank you for having me again.
00:05So maybe for people who are not familiar,
00:07tell us a little bit about what distinguishes Zooks
00:09from some of the other AVs on the market.
00:11And do you agree that they look like toasters?
00:15Toasters are cute.
00:16Computers like symmetry.
00:17So we are totally OK owning that.
00:20Look, I mean, at Zooks, we are reimagining mobility on demand
00:25with basically a purpose-built robotaxi.
00:29And since inception, we've been super consistent.
00:32We think that the purpose-built robotaxi
00:34is the best opportunity for safety
00:37and for rider experiences.
00:39And so that's what we've been up to.
00:41And since the last time, they are now on public roads.
00:43And friends and family are getting to enjoy it.
00:46And trusted riders are coming soon, too.
00:49Yeah, so purpose-built robotaxi, what
00:51does that mean you're not doing that some other AVs are doing?
00:54We are, well, a couple of things.
00:56First of all, we are not taking an existing passenger car
01:00and adding sensors and compute to make
01:03it essentially driverless.
01:06We know that works.
01:07I mean, we can see them in San Francisco.
01:09But we think that when it comes to rider experiences,
01:12when it comes to safety, when it comes to operational costs
01:16and the path to profitability, our approach is better.
01:19We're not saying anybody else's is bad.
01:22The other thing we are not doing is, I'm really sorry,
01:24we're not making your passenger car, whatever you drive,
01:27driverless.
01:28We're not in that business.
01:29Yeah, talk a little bit more about that.
01:31What gives you a better path to profitability
01:34than some other firms when you're doing
01:36this robotaxi-specific model?
01:38So what's great about Zoox is, from the beginning,
01:42they didn't go, oh, OK, there is self-driving technology.
01:45What do we do with it?
01:47It was more around personal transportation,
01:50around a better way to do mobility on demand.
01:53And it wasn't like, let's take a car
01:56and see what we can do to it.
01:57It was, in order to provide that and in order
02:00to have an electric fleet out there that is constantly
02:03picking up, dropping off, what is the best way to do that?
02:07And from there, we sat down.
02:10Well, I wasn't there, but the two co-founders
02:12and the founding team sat down and really,
02:16on a clean sheet of paper, looked at the architecture,
02:19looked at the design, whether it's redundancy, safety
02:22innovation, everything that was needed to be done,
02:24and really thought about the customer.
02:26We have pictures, actually.
02:27We have carriage seating, which now we take for granted
02:30because we ride every day.
02:32But we have a picture, I think it's maybe 2014 or 2015,
02:35where they put up two little banks.
02:38And they are sitting down and saying, OK, what's the spacing?
02:41What's going to feel intimate if you're
02:42traveling as a group?
02:44But what's also going to feel just spacious and comfortable,
02:47irrespective of what you essentially look like?
02:50So it's all of these mini little decisions
02:54that you make for the rider and also for safety.
02:59And that gives you a better path to profitability?
03:02Well, yeah, because that means you think about operations.
03:07I always take the example of aviation.
03:10It took a while to get to where we are now today.
03:13And a lot of thinking about the fleet management,
03:16thinking about charging, thinking
03:18about what is happening when you're roaming around, picking
03:21up customers, thinking about the layer up top,
03:23and frankly, also thinking about just,
03:26if we end up with a robotaxi where you still
03:28have as many humans taking care of it to get it to work,
03:32that doesn't really work.
03:33Your ratio of folks that are helping the robotaxi
03:37has to be good.
03:38Otherwise, you didn't really solve anything
03:40from an economic standpoint.
03:41Interesting.
03:42So as Andrew mentioned, you've recently deployed in Las Vegas.
03:46What have you learned from those AVs on the road?
03:49Well, first, I'm learning to sleep again.
03:52I'll be very honest and candid.
03:55The first week, I was like, wow, this is happening.
03:58Before that, we were in Foster City.
04:00And I get to see it every day.
04:03I pull up to work.
04:05I'm leaving.
04:06I'm in a meeting.
04:06I turn around, and I see it.
04:08So it's kind of like your little baby is right there.
04:10You can see it.
04:11Now you're sitting in the Bay Area.
04:14And basically, the ops team said, stop calling us.
04:17Because I was every day saying, how is it going?
04:21How is she doing?
04:22Anything happen?
04:23Was there an interesting scenario?
04:25And we've learned a lot also about how people are reacting
04:28when they see it, how to interact
04:31with all of the community.
04:34So lots of little pickups.
04:36And the riders are giving us feedback, including
04:38my own children.
04:39Yeah.
04:40Interesting.
04:41What do your kids say?
04:43They gave me a four out of five.
04:45Not so bad.
04:46No, I wanted a five.
04:47What didn't they like?
04:49A couple of things on a couple placements of things
04:53that they were like, you could have done better there.
04:56But you know, they are teenagers.
04:59It's nice to have an in-house focus group.
05:02So you mentioned scenarios.
05:04So what are the toughest scenarios for your vehicles,
05:07the ones that you are really still working on,
05:09maybe having a hard time cracking?
05:10Turning left, pulling over, snow?
05:13Oh, we're not working on snow.
05:15I mean, the vehicle is capable.
05:17We've tested it on snow.
05:18So from just the dynamics, that's fine.
05:22But in terms of the business itself and the AI stack,
05:26we are not working on snow.
05:28Look, nominally, we do quite well.
05:31But it's every little, things went a little faster.
05:35Things went a little slower.
05:36Somebody stepped in.
05:39Scooters, bikers, bike lanes, motorcycles,
05:42all these kinds of things.
05:45But we build a bank of scenarios around them.
05:48We make sure we build up our simulation.
05:50And one of the things at Zoox is we
05:51believe in tiny, small, medium, large.
05:53This is not something that's going
05:54to happen overnight, despite whatever was said in 2017.
05:58And so we're just building and building and making sure
06:00that we have a safety case that we can rely on,
06:03where all of this data and all of the quantification
06:06makes a lot of sense and is predictable and repeatable.
06:10Yeah.
06:10Well, a few months ago, there was an incident
06:12where regulators started looking into two situations
06:15where Zoox vehicles braked and then were rear-ended.
06:18What happened there?
06:18And did you make any changes following those incidents?
06:21Yeah.
06:21So we have a test fleet that has essentially
06:24the identical compute and sensor configuration.
06:29If you are in San Francisco or Las Vegas or Seattle,
06:32you see them roaming around.
06:34We're also starting to test in Miami and Austin.
06:36So those test vehicles have a safety driver.
06:39They are in autonomy most of the time.
06:41But the stack is in charge.
06:43And the safety driver can take over
06:44if either they are not comfortable
06:47or if we're encountering a situation that we're not
06:49comfortable with.
06:50In this case, essentially what happened first,
06:53you collaborate and cooperate with the authorities.
06:56They are doing their job.
06:58We are sending out these machines and computers on wheel
07:02to drive amongst humans.
07:04So we welcome the collaboration, the participation.
07:08Second, we gave them all the information and data.
07:10Obviously, we looked at the scenarios themselves and said,
07:13is there anything we can do better?
07:16Is there anything we can do differently?
07:18And we've been doing that as we speak.
07:20And then we presented that information
07:22and essentially reloaded the software releases.
07:25Interesting.
07:26Well, there's also been a few other incidents
07:28from some of your competitors recently.
07:29The Federal Highway Agency investigated Waymo vehicles
07:32earlier this year.
07:33My colleague Jessica Matthews at Fortune
07:35published an investigation into an incident
07:38in which a cruise vehicle dragged a woman for 20 feet
07:40as it tried to pull over.
07:42So what is going on?
07:43And are these incidents leading the public
07:45to lose confidence in AVs?
07:48Well, there was a lot in your question.
07:50So we're going to back up the boat a little bit.
07:53First of all, I don't know that we're competitors.
07:55We're fellow travelers.
07:57In 2019, I remember looking at the California DMV test list.
08:03And I think there were 70 of us testing in San Francisco.
08:07I think the industry has matured and consolidated.
08:10It's a huge, very interesting market.
08:13And there's a handful of us that are going after it,
08:15which is good.
08:17You're never going to hear me say a competitor or fellow
08:20traveler did something bad.
08:22That's just not going to happen.
08:23Along the way, and in deploying this technology,
08:26there will be issues.
08:27The question is, how do you handle them?
08:29How do you deal with the authorities?
08:32How transparent are you?
08:34And making sure that you talk to the public overnight.
08:38Now, if you show up with a gigantic fleet
08:41before you're ready, bad things will happen.
08:44But yeah, the authorities will deal with you,
08:46and the customers will deal with you.
08:48When it comes to trust, I don't know
08:51that we were in the trust business, but it's implicit.
08:54It's expected.
08:55So we have to deliver it, and we have to deliver value.
08:57If we don't do that, customers won't ride.
09:00And if they don't ride, we can't sustain a business.
09:03But if we do it in little steps, and we take feedback,
09:07and we're transparent, we'll be just fine.
09:10So what do you think customers are saying right now?
09:13How are they feeling about AVs in this moment?
09:15From what I'm reading online, and also
09:19what we're seeing with folks when they ride,
09:22I hope you come by and take a ride.
09:24In Las Vegas, specifically, it's pretty impressive
09:27in terms of everything that's happening.
09:29Look, there's a little bit of apprehension.
09:35I mean, there are no manual controls.
09:38It feels like a living room that's transporting you.
09:41But then they get in, and there's a lot of excitement.
09:43They are looking around, pushing buttons.
09:45Obviously, you have to buckle up.
09:47If you don't, by the way, you will get a little message
09:49telling you to buckle up.
09:51And then you push Start, and you go.
09:52So the first two, three minutes, it's like, oh my gosh,
09:55it's going.
09:56And there's a lot of looking around.
09:57People are taking pictures.
09:58I have to tell people, you're not the celebrity.
10:00The robotaxi is.
10:02And then people forget.
10:05And they start discussing, and talking, or reading,
10:09or doing whatever, reclaiming that space,
10:11and doing whatever they want.
10:12So that's a general sentiment.
10:13A lot of wonder, but a lot of, oh, this is normal.
10:16I can see this being my mode of transportation in the future.
10:20Interesting.
10:21I'm going to come to everyone for questions in a moment.
10:22So if you have anything, please start thinking.
10:25Aisha, just in general, are robotaxis
10:27safer than human drivers?
10:30Robotaxis that are deployed on public road
10:32are safer than human drivers within the operational design
10:37domain they are deployed in.
10:39They are not ubiquitously safer than human drivers
10:43everywhere, which is why tiny, small, medium, large,
10:45and which is why you stay within your ODD,
10:48because that's where it's proven.
10:49Having said that, I want to be very clear.
10:52We talk a lot about the fatalities
10:55in the United States, somewhere around 43,000 a year.
10:59And it is true that the majority of those
11:01are caused by human drivers doing something
11:04they shouldn't be doing.
11:05And computers won't do that.
11:07However, I think it's really important
11:09to also say that, collectively, in the United States,
11:13we drive 100 million miles per single fatality.
11:18That is an incredible bar to meet.
11:20And so there are a lot of sort of,
11:23nominally, everything goes well.
11:25But you have to make sure that the unpredictable,
11:28you are able to handle.
11:31OK, any questions for Aisha?
11:34Oh, one right there.
11:35And please introduce yourself.
11:38Hi, my name's Jessica Matthews.
11:39I'm a reporter at Fortune magazine.
11:42One of the reasons that some of your competitors
11:44have left steering wheels and the pedals in the vehicles
11:48is in case an incident happens and a human
11:50has to come into the car and take control of it.
11:55And Cruze has been developing Cruze Origins.
11:58They've been slow to release them on the road.
12:00Those also do not have steering wheels and pedals.
12:02So I was curious, what exactly happens with your vehicles
12:09if a human does have to take over
12:11and they are not able to manually drive the vehicle?
12:15Thank you for your question.
12:16Much appreciated.
12:17You definitely follow the industry.
12:20This vehicle, there is no taking over manually.
12:25So I think your question is, if it doesn't
12:28know how to do something, it calls the teleguidance center
12:32and says, I need help, either for permission
12:35to break a rule safely.
12:38AI is still in charge, still making the decisions.
12:40Or we give it breadcrumbs, we call it,
12:43for a suggested trajectory that it still decides.
12:46So let's say a very sort of unstructured construction zone,
12:50and then it is still in charge after it
12:53gets a trajectory of saying, I can make that trajectory or not.
12:56Now, there are situations.
12:58We don't just look at, when it comes to safety,
13:02we look at the scenarios, but we also look at not being stuck.
13:07And so we put an incredible amount
13:09of work in not being stuck prior to deployment,
13:12because we understand that.
13:14Now, if you're stuck, depending on the level of being stuck,
13:17like you can't get out of it at all,
13:19that's the worst case scenario, because then you
13:20have to send a tow truck.
13:22But obviously, we work, and we do all of our simulation
13:25and all of our quantification to make sure
13:28that those are very rare.
13:30And in between, we can give commands to the vehicle
13:33to pull over or to take itself to a certain destination
13:37and how much time it has to do that.
13:39So there's a whole level of thinking
13:41that is blended in the deployment
13:43exactly for those reasons.
13:44But no manual.
13:46I mean, the way we look at life is,
13:48if you have to count manual interventions,
13:51this is not a robo-taxi.
13:54Interesting.
13:54Any other questions for Aisha?
13:55Oh, one right here.
13:58Hi there.
13:58Jonathan Foster from Zenith Capital.
14:00I'm curious if you think that the current system of cameras
14:05and computer vision on your vehicles
14:08is sufficient to take this industry all the way
14:10to a mature, safe environment?
14:13Or will we need some kind of federal and state
14:16infrastructure to kind of cabin what
14:19these companies are building on, like a network of sensors
14:23or other types of guide rails for these vehicles to operate
14:26in?
14:27Great question.
14:28So this is a little bit of a philosophy discussion
14:33or posture.
14:34At Zoox, we believe that we have to be
14:37able to deploy the robo-taxi within
14:39the current infrastructure.
14:41Why?
14:42Because when it's a safety-critical product,
14:45you don't want to count on infrastructure
14:47that might be available in LA but not in Atlanta,
14:50or might be available or calibrated differently.
14:53So we built and architected our sensor suite
14:57and compute with the understanding
14:59that we live within the current infrastructure.
15:02Now, in the future, if infrastructure becomes
15:05available, we have a very modular architecture
15:08that will allow us to take advantage of that.
15:11But out of the gate, that is not something we want to rely on.
15:14Got it.
15:15Another question over here.
15:17Hey, good morning.
15:18Todd Bennett, Deer Valley Resort.
15:19When you talked about not yet being in winter,
15:22my heart sunk just a little bit.
15:24I'm so sorry.
15:25Two-part question.
15:26One, could you talk about the operational challenges
15:28in winter?
15:28They're probably obvious, but I'd
15:29love to hear from your perspective.
15:31And number two, would you ever be interested in a pilot?
15:35OK, I'll start with number two.
15:37Let's talk.
15:38And then, look, the thing we pride ourselves with,
15:43this is a very serious endeavor.
15:46And I think you've seen that if you do a lot of hype, hype,
15:48hype, eventually the chickens, they come home to roost.
15:52So we don't want to be in that situation.
15:54The issue with snow is that, at least today, we use maps,
16:00and we use what we call our view of the road network,
16:04meaning where things are.
16:05And when it's snowing, again, the dynamics of the vehicle
16:09are fine.
16:11We've been testing in snow for a while.
16:13The problem is the world is changing constantly.
16:17And depending on which plow driver was, he or she or they
16:23plowed differently.
16:24And where we thought there was a road, there isn't one.
16:27And if it's happening on just, let's say, a short distance,
16:30that's not a problem.
16:31But if it's continuously happening,
16:34there are things that we need to do differently in perception
16:36and in planner, meaning trajectory generation,
16:39that we're not focused on right now.
16:42That's the honest to God truth, at least at Zoox.
16:46That's tough.
16:46Well, I'm in New York, and I feel
16:48like it's a similar version of the problem.
16:49Well, New York has lots.
16:50I mean, it is the Holy Grail for the business, but whoosh,
16:54not easy.
16:56That's for sure.
16:57All right, so let's talk about commercialization.
16:59When do you think the public will allow for it?
17:02I think, do you mean for the industry,
17:04or do you mean for Zoox?
17:06Maybe both.
17:06Both?
17:07I mean, the public is already allowing it.
17:09If you are around San Francisco or in certain parts of Arizona,
17:16it is impossible to not see Al Saeed, Awemo, Ye, Takedra,
17:21and Dimitri roaming around.
17:23And it's open to the public, by the way,
17:25so there's no special, you don't have to have a connection.
17:28You can literally go to the App Store, download the app,
17:30and ride, and experience it.
17:33For us, we'll start that journey end of the year,
17:36beginning of next year.
17:37Right now, we have external people riding,
17:39but they are not paying customers,
17:41again, giving us a lot of feedback.
17:42So if we continue doing that, tiny, small, medium, large,
17:46we engage with the community, we provide the value,
17:49and we don't do anything stupid as an industry,
17:52the public will, I mean, look, aviation, right?
17:55Occasionally, something bad happens, yeah?
17:58But we still go to the airport and get on planes.
18:01Well, we haven't mentioned yet Amazon.
18:03You're owned by Amazon.
18:05What kind of timeline does Amazon leadership
18:06want on commercialization?
18:08Whatever timeline we agreed upon as a plan of record,
18:11so the same timeline I just gave you, and then we build on that.
18:14We don't have a set of books for Zoox
18:16and a set of books for Amazon.
18:17It's so tiring to remember two different stories,
18:19because that's just not how we roll.
18:21They have the same view and the same plan I just gave you.
18:24Yeah, OK.
18:26So let's do some predictions.
18:30AVs, Zoox, five years, 10 years, what does it look like?
18:37Five years, 10 years, we are doing well in the Sun Belt.
18:41We are starting to figure out how to drive where,
18:47I mean, we can handle rain, fog, and everything,
18:49but snow is starting to be solved.
18:51Highways also are more ubiquitous.
18:54And then a surprise, I think we'll see a time expansion.
18:59There are a lot of folks right now who don't ride,
19:02but I think they'll be very attracted
19:03to this form of mobility on demand.
19:06Yeah, interesting.
19:07And you've mentioned aviation a few times.
19:09You're on the board of Joby Aviation.
19:11So you have insight into the future of transportation
19:13and mobility beyond just robo-taxis.
19:15So if you kind of expand that, how are we
19:17going to get from point A to point B
19:19in any mode of transportation in five or 10 years?
19:22What we're all hoping, I mean, this
19:24is the United States of America.
19:26It was built a lot on automotives,
19:28so we're not thinking it's going to be zero cars.
19:31But we're hoping to start convincing some of us
19:35to not have 2.3 cars, maybe have one, one point something.
19:39And when you're in dense urban environment
19:43or in environments where it doesn't really
19:45make sense to be driving, you'll
19:47be using mobility on demand, fully electric,
19:50fully autonomous, a cabin space that you
19:53can use for conference call, you can use for music,
19:56you can use for sleeping, whatever rocks your boat.
19:59And it's an extension of your space.
20:01And then when you're not using it, somebody else
20:03is using it.
20:04I mean, that is the way to generate scale,
20:07to generate value for customers, for society.
20:10And we hope that that's something we can deliver.
20:12Interesting.
20:13And just to circle back New York as the Holy Grail,
20:16what makes New York the most challenging place?
20:19Well, first of all, I mean, I don't
20:21think New York customers will enjoy the, hi, sorry,
20:26Zoox is not available today.
20:28Because it's snowing.
20:29That's just not how it's going to work.
20:31Second of all, Manhattan is tough,
20:34I mean, in terms of scenarios.
20:36On the one hand, it's a grid, which is nice.
20:38But on the other hand, it's one of the toughest traffic areas.
20:42And then second also, from a regulatory standpoint,
20:46there's still some work to do.