Promise of AV Technology to Transform Urban Mobility: Jeff Bleich on Cruise’s New Services in San Francisco


With the global autonomous vehicle market expected to reach a value of $2,161 billion in 2030, up from $76 billion in 2020, there is no denying that companies and governments expect AVs to bring tremendous change to our cities. To share insights into the current state of the AV market in the US and California, VX News spoke with Jeff Bleich, Chief Legal Officer and Chief Risk Officer for Cruise. In this interview, Bleich talks about what GM’s massive upcoming investment will look like, the current regulatory environment, and the technology’s potential to transform urban mobility.

GM announced this year that it was planning to invest $2 billion on Cruise in 2022. What leads GM to make that investment, and how will it be deployed by Cruise?

Jeff Bleich: The investment is mostly in our people. That’s been the secret sauce that Cruise has used. Find the best people you can in this space and allow them to do what they do best. It’s about financial incentives for bringing in the talent that’s required to do something revolutionary in the automotive space. Something that hasn’t been done in the automotive world for a century.

There are two reasons they’ve done it. One is that everyone was making a lot of little bets in this space initially, trying to figure out who is going to emerge and whose technology will work. Over time, we earned the confidence of our investors, and they really have wanted to double down. The second thing is a recognition that the status quo is no longer going to be acceptable around the world. There are too many players working too hard to move us to an AV/EV environment.  It is possible for people to go where we need to go in a way that people don’t have to die, and the planet doesn’t have to suffer. That’s our future and they want to grab it.

It appears that Cruise has concentrated on the City of San Francisco. With rules and regulations for autonomous vehicles emanating from the CPUC, which recently-issued permits to enable Cruise taxi rides on the streets of San Francisco between 10pm and 6am – how are local transportation agencies and the public adjusting to having AV’s on busy local streets and in neighborhoods?

We chose San Francisco very deliberately because the way machines learn is different from human beings. Human beings crawl, walk, then run. With autonomous vehicles, they can learn all those things simultaneously, but you need to have mitigations in place. Instead of starting in parking lots and then working your way up to cities, we wanted to go right into the city and find the most complex driving environment we could imagine. San Francisco really delivers on complex driving environments. We watch the videos the next day and see the assortment of things that our vehicles encountered, and we know we picked the right spot.

We also wanted a very attentive, engaged regulatory environment. What you want to do is learn what regulators are going to care about most and be able to address their needs, as opposed to just building what you’re going to build and thinking you’re going to convince regulators to be happy with it. It’s got to be a mutual engagement from the very beginning.

Many states don’t even regulate AVs, but California does. San Francisco even has additional layers on top of the DMV, CPUC, and the other state regulators. We wanted a tough environment and tougher regulators as a way of building this program out correctly. That’s why this announcement from the CPUC is so gratifying.

Obviously, there’s still work to be done. It’s a draft resolution, and we’ll go through public comment. We expect, probably in June, there will be a final hearing by the CPUC. All indications from them were what you’d want to hear; great confidence in us. We will be the first company, if the CPUC approves it, to be authorized to run a fully driverless ride hail passenger service in California and any major city in the United States. It’s a big deal.

In that vein, the DOT, at the federal level, has just changed some of its rules to make relevant legislative language inclusive for driverless vehicles. What challenges remain federally to enable more AVs on the road?

It's a combination. This is a tech race and a trust race. To obtain federal approvals, we need to stay cautious and patient enough that, even though our vehicles can do mindboggling things, we don’t rush out there too quickly. We’ve been very deliberate in how we expanded. As you mentioned, we’re going to start our ride hail service between ten at night and six in the morning. Both because it’s a time that people need a lot of ride hail, but also, it’s a less complex environment.

You want to build trust at the same time that you are bringing out your technology. We evaluate very rigorously how our vehicles are performing. We’re constantly doing additional testing with supervised vehicles at the same time, then transferring that learning to the driverless vehicles. It’s that process which keeps improving the technology every single day, but also builds trust. The first time you’re standing on a sidewalk and a ghost vehicle with no one inside it comes rolling up to you and stops to pick you up, you want to know that a lot of thought and care has gone into it before you take that first ride. What you discover is that two minutes later, you notice how good of a ride it is and you start thinking about other stuff.

Recently a local news outlet reported on an autonomous Cruise car’s encounter with police in San Francisco. Give us the context for that encounter and how the company learned from that event? 

One of our vehicles, named “Rigatoni” by one of our AVTOs, was driving in San Francisco. A technician who was responsible for looking after the vehicle decided to try and save some battery by turning the AV lights off while he was charging it, but then forgot to turn the lights back on. When the vehicle went out without its lights on,  a police officer recognized this. He did the right thing and turned on his siren, and Rigatoni did the right thing which was to think, “Okay. This police car must be chasing someone, so I need to stop in lane.” The police parked  behind Rigatoni.  But as the officer came over to take a look at the vehicle, Rigatoni realizes that it was actually the car getting stopped. So, Rigatoni did what it was trained to do. He’s got to go to a safe spot because that’s what the laws require. Once the officer was clear of the vehicle, he pulls away from the police and goes through the intersection to the nearest safe location to pull over.  But all that people who are watching and filming this see is the police coming up to a vehicle and suddenly the vehicle races off!      

All the late night comedians had a lot of fun with this. Trevor Noah said that he loved the vehicle’s confidence.  He joked that it operates with the confidence of a white man.

What I liked about the videos was listening to the commentary of people who are on the street watching us for the first time and asking each other what was going to happen. How do the police actually cite a vehicle or communicate with it? How are we going to negotiate with these vehicles? It was nice for me to actually see people asking these questions, being curious, and also looking at Rigatoni as a student driver, feeling some empathy for the police and for the vehicle in this situation.

We do have a whole set of protocols that allow police to communicate with us. We’ve met with them and made sure that everyone’s trained on them. It was good, though, for us to understand how this can play out in the real world. It raised people’s consciousness about the fact that AVs are coming and coming fast, and the types of adjustments that we’ll all be making.

Cruise recently launched driverless Robotaxi services in San Francisco, and has been working with Walmart to deliver groceries in the Phoenix area. Elaborate on the promise of AVs to deliver such services at scale in like cities.

We have become used to this bizarre status quo where we just accept that to get from one place to another, every year, about 40,000 people are going to die and another 4 million are going to be injured. There’s an idea that somehow that’s the necessary price of mobility.

It just doesn’t have to be that way. The least reliable feature of every vehicle is a human driver. It’s because we’re human and we get distracted, get drunk, and our capacities tend to diminish over time. We have all sorts of limitations in how we operate a vehicle.

We can create a vehicle that doesn’t require our intervention and can do all the things we need to while continuously learning. When one human driver learns how to drive better, it doesn’t help anyone else in terms of their driving ability. When you train a vehicle to drive better, that knowledge is immediately transmitted to every other vehicle. If you do this right, over time, you’re going to have much more efficient and safer vehicles.

There are going to continue to be challenges on our roads, but over time, I think people will find that they actually like having an AV better. Instead of a ride hail, where you’re in that awkward situation where you might need to talk to a stranger who is driving, you basically have your own vehicle. It’s all to yourself, but you don’t worry about paying for insurance, registration, gasoline, maintenance, and parking. It doesn’t depreciate the day you take your first ride in it. Instead, you get all the benefits of mobility, but it’s chauffeured and there are no burdens associated with it.

Such services require that the autonomous vehicle operator in any urban jurisdictions to have mapped the city involved in great detail. Cruise has had an opportunity over years to do such mapping in San Francisco, but what must now be accomplished— and how long will it typically take—for Cruise to provide like services in other cities?

You’re exactly right. We start with mapping. We’ve gotten very good at being able to do that efficiently in a granular detail. On top of that, we focus not just on the features of the environment in which we’re driving, but also what are the unique ones. If you’ve been driving in San Francisco, for example, you probably haven’t encountered a lot of snow. You’ve got to start addressing those different features. If you’re driving in an area with lots of glass buildings and sunlight, you’re going to be relying more on your LIDAR than on your cameras. Then you do another round where you’re adapting the training of the vehicles to the unique features that they haven’t already learned in other driving environments.

From there, you roll out operations in a build-up approach. For example, we’ve been asked to be the exclusive Robotaxi provider in Dubai. This is what we’re doing there now.

Pivoting back to your legal responsibilities at Cruise, to the extent you are permitted, share what those responsibilities include and how they’re evolving as new policies are adopted by local, state and the federal government.

If you’re a lawyer who really loves the law and policy, this is a dream job. We have to work alongside our regulators, together, because there’s no playbook. If you think about when vehicles first appeared on the scene at scale over 100 years ago, there weren’t traffic lights, lines on the road, and people didn’t get driver's licenses. Lawyers, regulators, and safety experts needed to invent these.

What we’re trying to do is learn and work with regulators on how these vehicles should be regulated. What features can government incorporate into local infrastructure that’s going to make this work better at the same time?

There are a lot of talented companies in this space, and there’s going to be competition. So, we’ve got to protect our IP. One of the things that we have recognized over time is the value of the tools we had to create to make these vehicles. There’s valuable IP that has lots of other applications beyond AVs. We expect that there’s going to be a risk of other companies using that IP without paying licensing fees or of some enterprising lawyers trying to bring actions claiming that somehow someone else developed this technology before we did. We’re preparing and we’ve worked very hard to ensure that we document all of the technology and innovation that we’re doing.

Product liability laws could be pretty different. With our vehicles, there’s no distinction between manufacturer liability and driver liability. In some ways, that means that there may be some real efficiencies that we can develop. On top of that, we have much greater fidelity into understanding what our vehicles are doing than what typically happens. Typically, there’s an incident on a road and everyone’s got a different story of what happened. We’ve got data and have recorded what happened in minute detail. That may have some dramatic changes in how liability is assessed and how efficiently these issues can be resolved.

As you well know, a number of candidates are running for office today on platforms to ban autonomous vehicles. There are others who endorse and want to incentivize AVs. Elaborate on how you are engaging presently with the public and elected officials.

As I said earlier, this is as much about building trust as it is building technology. I have been meeting with everyone who might have a hand in regulating the industry to let them know what the real challenges are and how we can work on them together.

There are a few things that you always hear, like people want to raise the trolley problem and how we're going to solve that. That’s not one of the big issues that AVs are going to confront. We don’t program a vehicle to hit this instead of that. What you do is train the vehicle to be able to simultaneously identify 1000s of different scenarios and then do whatever it takes to avoid any collision. There isn’t a binary choice between hitting this car and that car. In many ways, it’s almost like a human reaction in the sense that you have to rely almost on instinct. The vehicle is developing the instincts of the very best drivers you can have.

Lastly, without triggering an SEC review, share what you believe the size of AV market will be in five or ten years?

In five or ten years time, I think we will be at a point where you’re seeing AVs in major cities. You’ll also see a number of AVs in the long haul trucking industry. We’re not doing that space at Cruise, but in some ways it’s a less complicated space to operate in.

I think people will be used to the idea of AVs. It will do something over the next 10 years that’s very similar to what happened with the move from horses to vehicles. It will feel like it’s going slowly at first, and then all at once, everyone needs to have one or be subscribed to one. I have the sense that AVs will just be everywhere in about ten years.

“…[There] is a recognition that the status quo is no longer going to be acceptable around the world…We will be the first company, if the CPUC approves it, to be authorized to run a fully driverless ride hail passenger service in California and any major city in the United States. It's a big deal.”