Cruise, a majority-owned subsidiary of General Motors, has secured funding from corporate juggernauts like Honda, Microsoft, Softbank, and Walmart.

In May this year, Cruise was the first of its kind to be granted a permit from the California Public Utilities Commission to give passengers driverless rides in prototype robotaxis. This marks a significant milestone for the organization that can now unleash its fleet of driverless Chevy Bolts onto the road to test their tech out in complex urban environments with real passengers.

However, the proper fun will start when their flagship vehicle—the aptly named Origin—hits the production line sometime in 2023. The Origin is the company’s first vehicle specifically designed to operate without a driver on board. Unlike the Chevy Bolts, it does not have manual controls such as pedals or a steering wheel.

So we decided to cruise on over for a video call to speak with the company’s Director of Research and Insights, Arianna McClain, to get a better look under the hood of the research process at one of the most exciting autonomous vehicle organizations in the world.

I was initially hired as the Director of User Experience Research. But over the course of the past nine months, I’ve started overseeing research across all users as the Director of Research and Insights. We’re in charge of research for engineers and engineering products. We have a hardware team, and a traditional ride-hail and delivery experience team, which is probably what most people think of when they think of User Experience Research at Cruise. We also have a new Autonomous Vehicle Behavior team working on understanding what riders want from autonomous vehicle behavior.

You have to realize self-driving cars have never existed before. And this is what you’re going to hear me say a lot, and this is what I say when I interview candidates—how do you do something that’s never existed before?

That means all the tools engineers use to develop self-driving cars have never existed before. Imagine having dozens of engineers working on one autonomous vehicle behavior, such as an unprotected left turn. These teams that make the brain of the car smarter need to collaborate and understand if their models are improving autonomous vehicle behavior over time. So our research team spends a lot of time with these engineers to understand if their tools are easy to onboard and use to speed up the development workflows and increase collaboration.

I can give you a very specific example of how we need to think about autonomous vehicle behavior.

Our autonomous vehicle behavior team is going to start thinking about certain things like pick-ups and drop-offs to try and understand where the safe spots to pick people up are. To do this, we will have a user researcher talk to riders and understand what a safe spot looks like. And then, they will work with our legal or policy team and our engineering team and tell them what riders want. We will also work with the city to pass on the information around the areas that are okay to drop off in that are also consistent with people’s needs.

We need to think of unique autonomous behaviors and what those autonomous behaviors might be. For example, how do riders expect our vehicles to react to an emergency vehicle? How should they respond regarding pick-up and drop-off? Should they go the speed of traffic if it’s over the speed limit? These are all the things we need to think about to create a human driving experience that doesn’t feel robotic.

It is amazing to partner with General Motors and Honda as both an investor and partner because it gives us the flexibility to operate as an independent startup but with big company benefits. I’m going to give you a few examples from a research perspective. When I first got to Cruise two and a half years ago, we weren’t ready to have passengers ride in our vehicles. The technology wasn’t developed. From a User Experience research perspective, we needed to understand (and we’re still trying to understand better) what information we need to provide in the UI or onboarding, for example, to make riders feel safe and comfortable.

To do that, we partnered with the super-talented and experienced R&D researchers at General Motors. One of my favorite stories to tell is about how GM loaned us a right-side driving vehicle, which sounds super basic, but it really helped. This meant we could hide real drivers and tell people there was technology hidden on the right side of the vehicle. We drove around the GM campus and had a trained driver act like a self-driving vehicle. Six out of the seven riders thought they were in a self-driving car. We had things like a fake app that would allow voice input from the passenger to control the vehicle.

GM also has a cutting-edge lab facility that enables us to conduct research. So we were able to run a few studies in their simulator. We also have the GM’s Milford Proving Ground to drive around the race tracks in autonomous mode. It’s very cool. I don’t know how other companies are doing it without these resources.

This isn’t like a traditional mobile web prototype that you’re asking people to walk through. We have to do Wizard of Oz prototyping or create paper and cardboard prototypes. We have to find ways to make the future feel like reality today because we can’t wait until our technology is ready to start developing the experience.

Our Leadership Team is very user-centered. We spend a lot of time and resources out in the field, talking to people, testing, and iterating. We’re trying to figure out what people need to know to feel safe and comfortable. We are putting people in scenarios that might feel unsafe (but aren’t actually unsafe).

We need to develop a product that’s not just technically safe but is perceived as safe and is exciting to be part of. We need to have our riders trust us before they get in, once they get in, and throughout the ride. This is something we’re constantly thinking about from a user experience and research perspective.

People often design for the happy path. We have to design for the unhappy path to avoid putting people in bad situations without guidance. They need to feel comfortable coming back and telling their friends about the experience. We don’t want this to be a ride that people take one time because it’s interesting. We want this to be how people get around.

We have to think about a lot of things. So, for example, we’re in the process of planning cross-cultural research for our upcoming launch in Dubai. We need to think about their basic assumptions regarding self-driving vehicles in this new market. This impacts onboarding, marketing, and practically everything else. For example, do we need to convince them it’s safe? What are the feelings of safety and perceptions of safety of autonomous vehicles in this market? We also have to think about cultural norms in the UAE. For example, we know that we have to think about font size and pixel size when designing the UI. Arabic is less concise than English. It’s wordier, which means it will take up more space on the screen and buttons. And so there’s a lot we need to think about from a cultural perspective. So, we’re going to immerse ourselves and talk to people so we can understand all the things that the research and design team need to consider when designing for culturally different markets.

When my team is writing their quarterly or annual roadmaps around what they’re focused on, I ensure that their goals level up to our OKRs. Because if it’s nice to have, are they going to do something with it? Or is it just nice to know? If you don’t know what someone is going to do with the work you are doing, then you shouldn’t be doing it.