Why I’m not interested in self-driving cars

Being in the tech world, it’s hard to go even one day without hearing self-driving cars mentioned. On Twitter, on the news, or in everyday conversations. Here in Chandler Arizona, you’ll even see a decent number of the Waymo cars driving around. It’s really easy to spot them because of how unusual they look: a white minivan with a huge siren looking thing mounted on top and a green W on each side.

Advantages to self-driving cars

If we pretend that fully self-driving cars are a reality (and are widely used) and think about the Pros and Cons, I’ll admit: there are some pretty cool Pros about self-driving cars

• You could save a lot of money if you didn’t need to own a car (no payments, insurance, or maintenance)
• Disabled or impaired drivers would have a safe means of transportation (handicapped, drunk, sleepy, etc)
• Software can react faster and more consistently than humans. This is great for safety
• No traffic jams if everyone had self-driving cars (efficient driving)
• You would never need to worry about parking; just call another car when needed
• If popular enough, Police could potentially focus less on traffic and more on serious crimes

These are some pretty nice benefits- I can definitely appreciate why folks want to solve this problem (and I’m sure I missed other great reasons). That said, I personally believe that it’s an overly complex problem which is not worth solving.

I like driving my car

I love cruising in a nice car. Windows down, breeze flowing, radio on with your favorite song- just relaxing. While I would say that I am a car enthusiast, I’m not an expert at driving by any means. I have only owned cars with an automatic transmission (I’ll grind or stall out on manual). And sure, there were times I’ve left my headlights on before until the battery ran down. But at the end of the day, I enjoy the experience. I am in full control over when and where I drive and I enjoy the pride of ownership that goes along with owning and maintaining a vehicle (for better or worse). I’d have to imagine there are many others like me that would not be willing to give up this level of control.

The human experience

Having a human driver pick you up and drive is something that software will never be able to replace. When I’m in a city I’ve never been to, I love being able to ask a driver about favorite places to go. They’ll also help load and unload your luggage. Maybe they know a route which has a really beautiful view of the ocean. Folks that are disabled or impaired and need help can count on the driver to help them get in and out of the car. Small talk can be super enjoyable sometimes- there are a lot of interesting people out there.

There’s also the reaction component that a human has. When driving, you might see someone on the side of the road that needs help. What if a crazy person gets out of their car with a baseball bat- how would the car know that you need to floor it and jump the median to escape? If someone has a heart attack in the back, the driver can drive them to the hospital or personally jump in and help if they have medical experience. Consider something as small as looking really sad and having the driver notice and put a smile on your face. There are just an unimaginable number of situations where I personally would want a human being there.

Complexity involved in solving the problem

This is the big one for me. Being a software engineer, I know from experience that focusing on edge-cases (situations that don’t normally happen) can eat up time and money. The cost of staffing engineers full time to solve a problem as large as self-driving cars is unmeasurably expensive. Whether people realize it or not, the folks working on solving this problem are working on artificial intelligence- it’s not just about reading sensors and reacting.

With today’s technology, the happy path (best case scenario) seems to be possible. But when you want a guaranteed 100% safe experience, you get into a lot of edge-cases. If the car will crash and you have to choose between hitting another car OR running over a person on the sidewalk, which do you choose? If a ladder accidentally falls off the back of a work-truck, do you run it over? or do you swerve into the next lane? How can you tell that other drivers on the road are high risk (ex: they have their lights off, or they’re clearly drunk / not paying attention) so that you can stay away from them? Should self-driving cars honk at people who aren’t paying attention when the light turns green? How do self-driving cars keep track of traffic laws, something that constantly change?

Conclusion

The amount of “what-ifs” and the effort needed to solve each of those problems is something that I personally believe makes the problem not worth solving. I enjoy the human experience associated with vehicles and I think it’s cheaper (and a better use of time) to continue to have a person drive. Moving towards a fully self-driving culture really seems to dehumanize the travel experience for me in a lot of ways.