AI & the future of safety

AICHA EVANS: We’re sending machines out there to drive among humans. People should ask us questions. People ask me all the time, “What’s waking you up at 3:00 a.m.? What’s stressing you out?” And I think they expect me to say, “I don’t know, capital or execution or this or that.” But I’m constantly asking myself, “Are we going as fast as we can, but as slow as necessary?” We just passed two million driverless miles on the robotaxi on U.S. public roads. That’s a huge milestone. And as you do that, you do become a household name because you earned it, not because you said so.

BOB SAFIAN: That’s Aicha Evans, CEO of the autonomous vehicle business Zoox. As robotaxis increasingly pop up across American cities, I wanted to hear about Zoox’s strategy as a subsidiary of Amazon, about its competition with Waymo, and about the plan behind a new Zoox partnership with Uber. Aicha talks with me about what it’ll take for robotaxis to go from consumer novelty to everyday routine, and how far down that path the industry actually is in 2026. She also shares lessons, both positive and negative, from her time at Intel and why she has what she calls an invisible army of rebels within the Zoox team. Aicha is direct, funny, and a beacon of optimism, so let’s get to it. I’m Bob Safian, and this is Rapid Response.

[THEME MUSIC]

I’m here with Aicha Evans, CEO of Zoox. Aicha, thanks for joining us.

EVANS: My pleasure. Thank you for having me.

Copy LinkWhy robotaxis are nearing a real proof point

SAFIAN: Zoox is in the red-hot center of building a new mobility future: electric autonomous vehicles. You have a new partnership with Uber. Your robotaxis are operational in Las Vegas and San Francisco. How close are we, really, to a dramatically different mobility paradigm?

EVANS: I think, as an industry, there’s been a lot of progress. We’re at the proof-point stage. Over the last 20 years, we’ve had a lot of “Oh, it’s happening tomorrow morning” and “Oh, it’s never going to happen.” We’re past that stage now. The proof points are there, for us and for fellow travelers. Now it’s a matter of starting to prepare for scale. But I’ve always been very consistent that this is not going to be like a consumer product where, all of a sudden, boom, 100 million people experience it. It’s going to be step by step, but we’re well on our way, which is really exciting.

SAFIAN: Your most well-known fellow traveler, Waymo, has chosen to retrofit existing cars. You guys have opted for purpose-built vehicles with a striking design. It’s got two benches facing each other. There are no driver controls, no steering wheel. It doesn’t really look like a car. Why make that choice?

EVANS: If AI is going to be doing the driving, it’s really about the customer experience and also about the best way to materialize this product. First, you have the safety aspect. In a regular passenger car that is architected for a human driver, the safest place to be is actually the front seat. For us, we were able to look at redundancy. We were able to look at our optimal sensor architecture so that we can see things, including occluded things. In Silicon Valley, sometimes maybe we think about the customer secondhand. Here, they thought about it firsthand and about the customer experience. It just does not feel like you’re in a car. What we’re seeing from folks, both people who ride and people in the communities where we ride, is curiosity. The first couple of minutes are, “Oh my gosh, what is this?” And then, “Wow, this makes so much sense.” Look, you’re not doing the driving, so why have a bunch of things that are involved in the driving?

SAFIAN: Calling it an AV still feels like a novelty. That’s what I hear from others who request these robo-rides. Do you have clarity about what it’s going to take for self-driving to become more routine? Is that about the design? Is it about the tech? Is it about the product? Or is it something else?

EVANS: Look, this is a big deal. These types of transformations don’t happen overnight. Look at aviation. It took a while, but now you’re at JFK and you’re like, “How can so many planes be coming and going?” So I think we’re going to have to continue to deploy, continue to put safety at the forefront, continue to engage in a dialogue with the regulators and the communities. And then maybe a decade from now, or maybe two, I don’t know for sure. I’m not in the prediction business. But at some point, we’ll be like, “Yeah, I still own a car, but for certain modes of transport, it just makes more sense to use a robotaxi.”

Copy LinkThe relationship between Amazon and Zoox

SAFIAN: You must be good at projecting in some ways, because you were part of the team that made Zoox a subsidiary of Amazon, which acquired it for $1.3 billion in 2020. How does Amazon help Zoox’s trajectory? Is it about financial resources, access to technology and AI? How is Zoox different than if it were on its own?

EVANS: Focus, focus, focus. The financial backing is important. A strong relationship with AWS also goes without saying: the compute. One of the things I love about Amazon is the plurality and multitude of businesses and industries it has been in. We forget Amazon started by selling books.

So they’ve seen a lot. They’ve experienced a lot. There’s a lot of pattern recognition. There’s a lot of customer obsession. So we get a lot of advice. When something’s going really well: “Can you do more of that?” When something is going poorly: “Why is that? And how are you looking at bottlenecks?” This summer, it’ll be six years, so we’re way past the dating phase. I’ve been on both sides of M&A at big companies, and I would say Amazon gets maybe an eight and a half out of 10.

SAFIAN: Eight and a half.

EVANS: Yes. I have the freedom I should have. I tell people all the time, it’s not like when you have a startup that is fully in the private sector with a board. I’ve been on the other side, where you have VCs, institutionals and independents. There’s always a decision-maker and a boss. Make your peace with it. And we’re sending machines out there to drive among humans. People should ask us questions, whether it’s the regulators or our bosses. It is the right thing to do, and I welcome it, and we are better for it.

SAFIAN: There are people who get nervous about the idea of robots taking over, not necessarily robots on the road, but robots in our lives.

EVANS: Absolutely. I have two teenagers, and I’m like, on the one hand, you’re so lucky because this is really an era of transformation in almost every sector because of AI. On the other side, I’m like, why did I have to have young adults, teenagers, during the transformation phase? Because there will be a lot of change. Having said that, I’m on team optimism. Humanity has been around for a couple thousand years. I know we take ourselves very seriously because we’re the ones alive right now, but there’s been a lot of change throughout those years, and somehow we’re still here. So I tend to think that, statistically, why should it be different this time? We’ll have a lot of hard conversations. We’ll have some hard transitions. But in general, things will be better, because that’s how it’s always been.

SAFIAN: With your kids, do you teach them to drive, or is driving a skill they’re not going to need in the world they’re going into?

EVANS: They are 18 and almost 20, so they did learn to drive. But with my son, it’s interesting. If he’s in the city, he’ll order a Zoox. If he’s around our house, he’ll order Waymo, which is totally OK, by the way. It’s a great product. I use it too. I need to sit down and ask him on what basis he’s making those decisions, and hopefully I like the answer.

Copy LinkInside Zoox’s partnership with Uber

SAFIAN: Your partnership with Uber, I’m curious how big a deal that is, because Uber also partners with Waymo. What makes that deal meaningful for you?

EVANS: They do partner with almost everybody, which is great. If I were in their shoes, I would probably do the same. Dara has taken rides in Zoox, and Uber totally gets the differentiated experience. For Zoox, for almost 12 years now, we’ve been so focused on building the tech and building the processes. Over the last couple of years, we’ve really started thinking about commercialization and how we’re going to do this. I don’t see this category as just taking share from whatever exists today. I actually see an expansion of the market.

As far as Zoox and Uber, for us, it’s about learning. It’s about experimenting. I’m pretty sure if you arrive in Las Vegas, maybe you know about Zoox, maybe you don’t. Now you’ll see them on the Strip and be like, “What is that? Oh my gosh.” But I’m pretty sure you know about Uber. So right there, that makes it worth it to Zoox. And some transportation has nothing to do with pleasure. Frankly, it has to do with being utilitarian. If we can help serve that together and scale faster, the experiment will have worked.

SAFIAN: You’re expanding into other cities. You’re testing in LA and Austin and Miami. How do you decide where to go? I’m not sure how soon I might see them here in New York. Are some cities tougher than others for AVs? How much of that is weather? How much of that is density? How much of that is regulation?

EVANS: A little bit of all of the above. Weather is important, no question about it. For example, for Zoox, snow is not really a short-term priority. We’re working on it, but that’s not a city we’re going to pick to start. Having at least some evidence that this is something the community welcomes, at least the seed of demand, matters. The regulatory environment matters. I’ve said publicly in the past that New York is the holy grail, but first it has to be legal to have a robotaxi there. So please, please, please tell the political and governmental infrastructure that you would like robotaxis in New York.

SAFIAN: Zoox isn’t yet quite a household name, but it’s kind of an old-timer in AV terms. You mentioned it’s been around for 12 years. Is that an advantage in AVs, or does the tech stack change so fast that building today could be just as fast? What are the pros and cons of that time?

EVANS: I think part of it is choice and leadership style and culture. Being a household name, to me, I believe in the earned life. Nobody has a God-given right to anything. You have to work hard at it. And as we’re out there serving more customers, we just passed two million miles. For us, that’s driverless miles on the robotaxi on US public roads. That’s a huge milestone. We have roughly 500,000 people on the waitlist. We have close to 400,000 riders who have given us feedback. We’re continuing to put points on the board. And as you do that, you do become a household name because you earned it, not because you said so.

Copy LinkHow generative AI is speeding up autonomous driving

SAFIAN: The explosive growth in generative AI: how does that change what you do and how the AV world is going to develop from here?

EVANS: We’re going faster. That’s really how it’s impacting us. Simulation is more capable. Our ability to correlate data and information is better. Engineers, actually, I shouldn’t even say engineers, every Zoox employee is way more productive. You get to answers quicker. Everybody’s talking about physical AI too. I’m like, “Wow, yeah, we’ve been doing physical AI since we were born.” Every time there’s one of these big transformations, a new set of giant companies is born, and some companies maybe don’t adapt and don’t figure out how to be the elephant that can dance, and they’re replaced. I expect basically the same dynamics, but bigger and probably faster than we’ve ever experienced.

SAFIAN: Do you think, “Oh, we should rebuild our stack from scratch because we have this new capability”?

EVANS: Not for us. We have a certain architecture we believe in. There are really big parts of the stack that we are able to modernize, and it’s not the first time we’ve modernized it, by the way. Sometimes I’m like, “Well, what if somebody looks at our code?” Then I’m like, “By the time they can understand it, it’ll have changed again.” Because we’re learning constantly, applying new techniques and new algorithms. We basically modernize our stack as we go. But it’s important to have explainability and traceability because, no matter what we do, the systems will be safer than humans, but they will also make mistakes. They will not be perfect. And being able to understand when a mistake was made, why that was the case, and how to correct it is really key. You can’t just have this part where you throw in a bunch of things and then a soup arrives and you can’t decompose the soup backward.

SAFIAN: I do love that term explainability, which, for many Silicon Valley companies, is a euphemism for, “We really don’t know how we got to these outputs, and we’re trying to trace it back and figure it out.”

EVANS: Yeah. At least in driving, I don’t like commenting on other people’s businesses because I don’t like it when people do it to me, because I have no idea what’s going on inside. But at least for physical AI, for driving these robotaxis in human communities, that is unacceptable. That’s our view.

SAFIAN: You have to know.

EVANS: You have to know. Period.

Copy LinkWhat China reveals about the future of EV and AV competition

SAFIAN: I talked with Ford CEO Jim Farley recently, and he was all amped up about China’s auto industry, the scale, the advances, the pace. How much do you look at China and what’s happening there?

EVANS: From an AV standpoint, our expectation is that there will be an ecosystem in China that we don’t have access to as an American company. Now, I think that when it comes to Europe, Middle East, Africa, LATAM and so on, you’ll probably have a mix of both ecosystems. So we keep track of them with that basic foundational premise. I think with EVs, as you go from internal combustion engines, or ICE, to EVs, you start going from a machine with wheels to a computer on wheels. And that means that integration between hardware and software, and not treating software just as basic control functions or as controlling the infotainment, but really having software be a first-class citizen in the architecture design and having decision rights, becomes very important.

Lots of people are very good at hardware. Lots of people are very good at software. Not a lot of people are good at both and treat them as having equal decision rights when it comes to major architectural and strategic decisions. The ecosystem in China, when you look at a lot of these consumer electronics companies that came from the smartphone era and that by definition had to be good at hardware and software and system-level thinking, is finding a glide path to EVs and AVs because that foundational knowledge and mindset are already native to them. And if you’re not native, then get busy being native.

SAFIAN: Aicha is definitely busy. She’s digging into the details of Zoox’s operations, managing Amazon oversight and trying to create a whole new future. So how does she use what she calls her invisible army inside Zoox to get things done, and where did the idea for an invisible army come from anyway? We’ll talk about that more after the break. Stay with us.

[AD BREAK]

Before the break, Zoox’s Aicha Evans talked about how she’s leveraging both Amazon and Uber to get more robotaxis on the streets in 2026. Now she talks about leveraging enthusiasm without slipping into hype, why she deploys what she calls an invisible army inside Zoox, and lessons from her tenure at Intel. Let’s dive back in. You clearly have a lot of enthusiasm about this area, and you alluded to this before, this balance between not extending too far into hype but maintaining that enthusiasm over a long period of time. How do you think about modulating those two things?

EVANS: I live by something that I was taught by a friend of mine, which is that you have to think about going as fast as possible, as slow as necessary. You also have to build a culture where you earn it, with a lot of transparency. Sometimes people visit and they’re like, “How do you know all these people?” I’m like, “I work at it.” Because I want to know that they all feel comfortable telling me if they think there’s a problem somewhere so that we can deal with it, because you have to earn it. If it’s a culture of fear, or if it’s a culture where you’re not making it easy and providing several channels to do that, you won’t go as fast as possible.

People ask me all the time, “What’s waking you up at 3:00 a.m.? What’s stressing you out?” And I think they expect me to say, “I don’t know, capital or execution or this or that.” But I’m constantly asking myself, “Are we going as fast as we can, but as slow as necessary?” Because we all know what happens when you go too fast and you don’t consider safety. Bad things happen, and they should happen, by the way. So the balance and the blend are the struggle and the triumph.

Copy LinkWhy moving fast still requires moving safely

SAFIAN: Prior to Zoox, you spent some years at Intel. Are there things from that experience that you draw on, or is Zoox such a different business that you look elsewhere for lessons and inspiration?

EVANS: Both. In general, I’m a curious person. I have this wonderful coach who taught me 15 years ago that it’s OK to ask for help. That is actually a sign of strength. So when I’m stuck, I’m known to pick up the phone and say, “Hi, I’m Aicha Evans from Zoox, an Amazon company. I need some help. So-and-so told me about you or introduced me to you.” But taking it back to Intel, I learned a ton there. I spent 12 years at Intel. It’s an important company. I root for that company to this day. I learned about hardware-software integration. I had a front-row seat to people who are hardware-only or software-only, and I was like, “Huh, that’s a problem.” I learned about process too. You’re going to laugh. If some ex-Intel or current Intel people who know me listen to this, they’re going to chuckle because I was known as a little bit of a rebel.

I complained about every process: “Why are we so slow and so dogmatic and so bureaucratic? Don’t you understand?” And yeah, well, guess what? Thank you. Thank you for that training, because then I came to Zoox, and it was still a very early start-up, a little less than 500 people. It had a lot of technology, but it really needed to be orchestrated and coordinated and to put processes in place, put orgs in place, put a common language in place, in order to succeed. So there are lots of learnings that I apply, but also lots of things not to do that I won’t tell you about.

SAFIAN: That you won’t tell me. Yes. Because you have to move as fast as you can as well, not just slow. I guess that’s what being the rebel in that group means. I guess you want to have your own rebels within your own organization, but maybe not too many of them.

EVANS: I want enough of them because you need a uniform distribution in all functions. One of the things at Zoox is that, because we’re vertically integrated, one day you’re talking to a traditional automotive engineer about chassis, battery, suspension, brakes and harnesses. The next day you’re talking to marketing about how much we want to emphasize safety, or not. So you want to make sure that, I call them affectionately, my invisible army, or the invisible army, is distributed across the corporation. But we also have to have a contract that we will debate, we will discuss, we will consider alternatives, but once we make a decision, we commit and we move, and we don’t revisit unless there’s evidence that assumptions were incorrect. Then we do a little bit of a feedback loop and keep moving forward.

Copy LinkLeading as an outsider with resilience and inclusion

SAFIAN: I talked to the leader of a big, well-known brand who was distraught about the way immigrants and other diverse populations and women are being made more vulnerable over the last year or so since the Trump administration, the second Trump administration, came in. But at the same time, she felt like, as a woman of color who was born in another country, it is not her place to talk about it. She just can’t. It makes her too much of a target. Obviously, I don’t want you to feel like you’re a target, but I’m wondering whether you have any thoughts or experiences about this, or how you manage that part of being who you are.

EVANS: My views pre- and post-this administration haven’t really changed. The person I admire the most is Marie Curie. Well, there are two of them: Marie Curie and Nelson Mandela, and I’ll explain why. Marie Curie: every time I’m like, “Yeah, of course I get disrespected or assumptions are made that are very annoying.” That happens on a daily basis. But I’ve basically made the decision that I was going to live my life happy and not miserable, and I am going to shut people up by doing the work. And when I succeed, it feels really good. And when I fail, I’m like, “What the hell happened, and what can I learn?” After a good cry, of course. Maybe that’s a very selfish way of living life, but that’s what I learned from Marie Curie. Nothing is to be feared. Everything needs to be understood so that we can fear less.

And on my worst days, I’m like, wow, she was an immigrant. She was a wife. She was in physics way back then. She was a mommy. She had it way worse than I do. And somehow she figured out how to thrive and get two Nobel Prizes, by the way, along the way. So let’s just understand why what’s happening is happening, what we can do about it, and also understand that progress is never fast and it’s also rarely linear. So that’s how I live my life. Then you have Nelson Mandela on the other side. I don’t understand how you’re in jail for 27 years, 27 years. That’s a long time. And by the way, it’s totally unjust by any reasonable standard. You become friends with the jailers, seriously. And then you come out of jail, and you don’t go on a revenge tour.

You go on an “OK, what have I learned and what do I understand, and how do I bring people together?” tour. So that’s how I look at the whole situation. I think there’s been a lot of progress. I also fundamentally believe that inclusion is more important than diversity, because if you do inclusion, you will get diversity. If you only do diversity, you will get some drama. And I think that’s sort of what’s happening. But it will all work out because it always does.

SAFIAN: This was great. Thank you so much for doing it.

EVANS: I really enjoyed the conversation.

SAFIAN: Listening to Aicha, I was reminded of the move fast and break things era of tech development that turned social media from a promising tool into something impersonal, often angry, sometimes addictive. Her admonition to move as fast as possible and as slow as necessary is a prudent mantra, not just for robotaxis, but for all of us in the age of AI. I also want to double-click on her encouragement that asking for help is a sign of strength, not weakness, especially when things are changing so fast that we need to rely on each other for perspective and support.

On Friday, we turn the tables by talking to the other side of Zoox’s recent partnership, Uber, sitting down with President Andrew Macdonald on how autonomous vehicles fit into their evolving business model, plus Uber’s surprising expansion into hotels and what gas prices are really doing to their driver community.