Robot vs. Chicken Farmer

On the differences between riding in self-driving cars and human-controlled taxis

We were turning off Polk Street in San Francisco’s Nob Hill neighborhood when I stopped scanning my email and asked my driver how his day was going. He didn’t answer. I laughed; there was no driver. I was two minutes into my first Waymo ride and I’d completely forgotten that I was sitting inside a robot.

There are self-driving cars cruising all around San Francisco. This is old news to some people and sort of shocking to others. But it’s normal in a few select cities and has been for a while. In 2014, I sat in the back of an early prototype of a self-driving vehicle made by what was then called Google X, the moonshot division of the advertising giant. Back then, a driver sat behind the wheel for safety reasons. An engineer rode shotgun with his laptop displaying how the car saw the world. The ride was fine. Safe but glitchy. It felt like a robot was driving.

The 2025 version is completely different. The cars are safe but not slow or overly deferential. At an intersection, when it’s your Waymo’s turn to go, the car really goes. This mild aggressiveness is safer, since an overly cautious driver tends to prompt irrational behavior by other drivers. You disrupt the natural flow of traffic if you’re too nice. Or too slow. I know that from experience. My 25 MPH average speed in the summer induces dangerous driving patterns in people visiting Martha’s Vineyard from Connecticut.

Shotgun and Safety

During two recent trips to San Francisco, I took seven Waymo rides, all downtown. I didn’t like riding shotgun; that made me anxious. Otherwise the experience was completely smooth. The car provides occasional warnings, and I liked those. At one point, when I was getting ready to climb out, the robot warned me that a cyclist was approaching. To test the accuracy of this alert, I opened my door anyway.

A man on a street bike slammed straight into the door, mangling his front tire and knocking over my organic Ethiopian single-source pour-over coffee from St. Franks. Luckily, the rider turned out to be my college friend Will.

We sure had a good laugh over that one!

Will’s getting better now, too.

Can you talk to the robots?

The Waymo greets you when you climb inside. I tried to have a conversation with the car, only to learn the vehicles turn off their microphones unless you’re trying to contact rider services and speak with a human. This was pitched as a benefit, and the announcement encouraged riders to sing out loud without risk of being recorded. Personally, though, I’d rather be recorded, since I kind of think I missed my calling as an indie country singer.

At some point, I imagine and/or hope you’ll be able to converse with your self-driving car. Installing an LLM with voice capabilities wouldn’t be that hard. And I’d like to engage in small talk, maybe ask about a building I spot through the window or whether the car knows of any good coffee shops near my destination.

But will that coffee recommendation be any good? AI systems learn about and experience the world differently than we do, so their behavior and decisions don’t always make sense. At one point, my Waymo picked me up a half-block from the entrance to my hotel, then proceeded to send us into traffic up and around a neighboring block to head itself in the right direction. A human driver would have suggested I walk half a block in the other direction and meet on the opposite corner, as that would have cut a solid ten minutes from our trip.

Simple and logical for us but not apparent to a robot.

Your Waymo doesn’t know what it’s like to sip an expensive pour-over coffee, and convince yourself that it’s delicious, and you’re a better person for drinking it, even though it kind of tastes like bathwater with berries. Your Waymo never felt the joy of knocking its college roommate off his bicycle.

What’s going to happen to drivers?

When I rode in Ubers while in San Francisco, I asked the drivers what they thought of these robotic taxis. Two of the four didn’t speak English. As a result, we weren’t able to connect. If you’re going to compete with robots for a job that involves human interaction, you should really be focusing on making those connections. Your humanity will give you an edge. Also, I should learn Spanish.

The two drivers I did converse with offered fascinating takes. The first told me the Waymos can have the city. He’s tired of picking up drunk and stoned passengers who act like idiots and disrespect both him and his car. This lines up with an argument technologists often use about robots moving into the workplace: they will take the jobs we don’t want.

This driver’s fear, though, is that the Waymos will slowly move out of the city and encroach on his business. And his big question, as we drove on a highway lined with billboards proclaiming the greatness of AI, is what’s going to happen when half our jobs are turned over to robots. “How are we going to buy the stuff these damn tech companies are selling us if we don’t get paid?”

Great question! My next driver was an entrepreneur from Mali. Back in Africa he grows papayas and mangos and has a thriving egg business thanks to his 3,000 chickens. We figured out that our kids are just about the same age, too. I was surprised to learn that he’s all for self-driving cars.

“Why?” I asked.

“I know I shouldn’t say that because I drive an Uber, and they are going to take my job sooner rather than later, but I hope it’s sooner because it pushes us to be better. It’s going to push me to evolve.”

Bill Nye vs. Vin Diesel

These cars are not perfect. They have made and will continue to make strange and occasionally dangerous, even fatal mistakes. Then again, I was nearly rear-ended by a young driver eating a breakfast sandwich and checking her phone a few weeks ago. Highways today seem overrun by people who’ve watched too many Vin Diesel Fast and Furious movies and not enough Bill Nye videos on classic Newtonian laws of motion. Human drivers are dangerous.

We have two visual sensors called eyes spaced close together, both aimed forward, and two ears to capture and triangulate the sources of sounds. Waymos have sensors everywhere. A laser scanner on the roof, plus one at each corner of the car. There are visual cameras and radar, too.

Plus, they’re deeply experienced. Each car has effectively driven 20 million miles on real roads and 10 billion miles in simulation. I’ve only driven about 360,000 miles in my life, and the effective number is probably much lower because I space out so often. Last August, I drove halfway across Pennsylvania without realizing it. Luckily, I was going in the right direction.

What next for robot cars?

So, when are we going to have these ourselves, driving us around our neighborhoods? That depends on your local climate, economics, insurance, and of course technology. The Waymos were everywhere during my visits to San Francisco because the weather was perfect and the traffic flow relatively normal and predictable. As Daniela and I explain in The Heart and the Chip, though, you’re not going to find self-driving taxis cruising through a snowstorm anytime soon.

Yet this feels inevitable. If we humans can steer a car in those conditions using our limited little biological eyes, I have to believe engineers will figure out how to build robots smart enough to use all those extra sensors to do the same. Maybe they’ll need new sensors. Or maybe better brains will be the trick. But we’ll get there.

Oh, and I forgot my favorite part about self-driving taxis. The biggest revelation, which I didn’t realize until I’d climbed out of my first ride.

You don’t have to tip a robot!

Thank you, as always, for reading. If you liked this story, click the heart below, so I know what’s working, or forward it along to a friend or colleague.