What does it take to make a self-driving car safer than a human driver? At Waymo, the answer lies in countless hours of testing, data-driven decision-making, and a deep commitment to safety. According to Waymo’s Safety Data Hub, their vehicles now have 84% fewer crashes with airbag deployment, 73% fewer injury-causing crashes, and 48% fewer police-reported crashes compared to human drivers in cities like Phoenix, San Francisco, Los Angeles, and Austin. But building safe self-driving technology is only part of the story. Waymo is also looking to the future through new partnerships—like a recent collaboration with Toyota aimed at combining Waymo’s autonomous driving system with Toyota’s globally popular vehicles. The goal: bring this technology to even more people and places. In this interview, we speak with Waymo’s Director of Product Management for Simulation and Evaluation, John Karmel, to learn how the company balances innovation and safety, the role simulation plays in testing the Waymo Driver, and what advice they have for students eager to build a career in tech and AI.

1. What exactly do you do at Waymo, and how does your role impact the company?

I am the director of product management for simulation and evaluation—that’s my official role. One of my key responsibilities is working on the safety aspects of our software. I lead product management for the Simulation and Evaluation engineering team.

Our main job is to test the software before it goes on the road to ensure it meets our safety benchmarks. Simulation is a virtual environment where we can conduct a lot more driving tests than we could in the real world. Real-world testing is limited—you can only collect so much data from test miles on actual streets or even with driverless (rider-only) vehicles.

In simulation, we can conduct orders of magnitude more testing and assess special cases—such as how a vehicle responds to children’s reactions—which we obviously wouldn’t test in the real world. These are what we call “long tail” or “corner cases,” and simulation allows us to test the full distribution of possible driving scenarios.

On the evaluation side, which is closely tied to simulation, we define and measure what makes a good driver. This is a very analytical and data-driven process. We establish benchmarks that the vehicle must meet before we can say it’s safe enough to operate on public roads.

2. How does Waymo balance innovation with safety?

That’s a great question. I actually don’t view innovation and safety as a trade-off. The whole motivation behind self-driving technology is to reduce injuries and deaths on the road. We’ve set very clear internal safety benchmarks, and achieving those is how we know we’ve met a high safety standard.

Today, external data shows that our vehicles are five to ten times safer than human drivers.

Historically, Waymo has taken a “slow and steady” approach. We’ve been working on this since 2009—over 15 years now. From the start, safety has been our top priority. By consistently maintaining high safety standards, we’ve built credibility and trust in the market.

For us, innovation is the mechanism that helps us be safer. It’s not about moving fast at the expense of safety; it’s about moving smart to enhance safety.

3. What sets Waymo apart from its competitors?

The market is constantly evolving, and different players take different approaches. One thing that sets us apart is our use of a combination of LIDAR, radar, and cameras. Other companies rely on different sensor suites, but this multi-modal approach gives us a robust view of the environment.

Another key differentiator is our commitment to letting the data speak for itself. We work with independent organizations, including insurance companies, to validate our safety claims. They assess our data using consistent and transparent methods.

While I can’t speak to every competitor’s internal processes, I can say that at Waymo, transparency and rigorous data evaluation are core to how we operate.

4. What are some of Waymo’s biggest challenges today?

Right now, we’re in a period of exponential growth. We’ve built strong foundations, and now we’re scaling rapidly. For example, the number of trips we’re providing and the cumulative miles we’re driving have grown about fivefold compared to last year.

A year ago, our main challenge was proving that our software could scale while maintaining our high safety standards. We’ve now demonstrated that by driving a significant number of safe miles.

Our current challenge is continuing to scale to more cities while providing the same high-quality, safe experience that we’ve achieved in places like San Francisco. We want to make this experience accessible across the U.S. and internationally.

5. What advancements in AI will most significantly impact autonomous driving in the near future?

Waymo has been an AI company from the beginning. While AI is a hot topic today, we’ve been applying it to move physical goods and vehicles for years.

Advancements in AI will help us in several ways. First, they’ll lower the cost of running AI models, which are currently tied to expensive hardware. This will allow us to further improve the quality of our driving.

AI advancements also enhance both the routine aspects of driving and those rare, long tail scenarios that are difficult to handle. We see continuous improvements, and we’re applying those advancements to make our vehicles even safer and more capable.

6. What are Waymo’s short-term and long-term goals?

Our overarching mission is to reduce deaths and injuries on the road. In the U.S. alone, there are about 40,000 traffic deaths per year, and globally, it’s around a million. Our goal – short term and long term – is to eliminate as many of these as possible.

In the short term, we’re expanding our ride-hailing service into more cities. Every vehicle we deploy is an opportunity to replace less safe driving with something we know is safer.

Long term, we’re focused on developing the Waymo Driver – a generalizable autonomous driver that can transport people and goods in a variety of scenarios. This will also improve accessibility for individuals who currently can’t get from point A to point B easily. We see this as a long-term opportunity to positively impact society.

7. A question from Stuart Hall students: How does Waymo handle vehicle charging, and do you see Waymo completely taking over the auto industry?

Charging:

In San Francisco, it’s less obvious where our vehicles charge. In cities like LA, you can more easily spot our charging stations – sometimes in small parking lots, other times in large depots where hundreds of cars autonomously navigate and charge. You can even find videos of this online – it’s fascinating to watch.

Auto industry:

We don’t aim to take over the entire auto industry. That’s not our intent, and it wouldn’t be a healthy approach to the ecosystem.

Self-driving technology is an important innovation for the automotive industry, and you’ll see we’ve partnered with OEMs like Toyota. For example, we’re working with Toyota in preparation for expanding to Tokyo.

More OEMs are developing Level 2 driver-assist capabilities, and we think our technology can complement and enhance those efforts. Over time, we’ll likely see more collaboration between companies like Waymo and traditional automakers.

8. What advice would you give to students interested in tech and AI?

There are many paths into tech and AI. We’re at the beginning of a wave of technological innovation that I believe is on par with the rise of the internet.

When I graduated in 2001, the internet was transforming the world, and I think AI will drive similar productivity gains.

There won’t be just one narrow path to enter this field. Yes, data science and software engineering are important, but there will also be huge opportunities in areas like UX/UI design, marketing, and human-AI interaction.

For example, look at how companies like OpenAI are thinking about new interface models with leaders like Jony Ive. As AI technology becomes more widespread, we’ll need creative thinkers from many backgrounds to shape how people interact with it. So no matter your skill set, there will likely be opportunities in this space.