© 2024 Public Radio East
Public Radio For Eastern North Carolina 89.3 WTEB New Bern 88.5 WZNB New Bern 91.5 WBJD Atlantic Beach 90.3 WKNS Kinston 88.5 WHYC Swan Quarter 89.9 W210CF Greenville
Play Live Radio
Next Up:
0:00 0:00
Available On Air Stations

First Person: Behind the wheel of a 'self-driving' Tesla

A Tesla car is displayed in a showroom at a Brooklyn Tesla dealership on April 4, 2017 in New York City. (Spencer Platt/Getty Images)
A Tesla car is displayed in a showroom at a Brooklyn Tesla dealership on April 4, 2017 in New York City. (Spencer Platt/Getty Images)

This is part of our hour on Tesla and the ethics of self-driving cars. Listen here.

Self-driving technology has come a long way in recent years, but it remains far from perfect.

In a podcast special, we share what it’s like behind the wheel when Tesla software is in action.

MEGHNA CHAKRABARTI: John Bernal is a former Tesla employee who helped train their beta software, and he now drives a Tesla Model 3 Long Range. John left the company in February, and since then he’s been focusing on his YouTube channel called AI Addict, where he posts videos of the full self-driving beta software in action.

JOHN BERNAL: It’s very exciting to have access to that software. And I love it, but I would have to say it’s definitely a lot more nerve-wracking to be driving a self-driving vehicle than it is to just drive yourself. Sometimes you might drive the vehicle, and it might be a flawless drive. I might drive 30, 40 miles and never give it a disengagement.

And I might go through very complex situations like construction, mountain winding roads, intersections, etc. But then at the same time, you might drive that same route again, and you might have maybe 10 or 20 disengagements. So you never know if the same route will always be good. And that’s what makes it nervous, because it does sometimes just jerk the wheel to the right. And you have to be there holding the wheel, paying attention, to take over, because it is a beta, after all.

BERNAL [Tape]: And there is a UPS truck in front of us. Just be interesting to see what happens.

[Car beeping].

I am taking over, I am trying. All right, there we go. Yeah, I don’t know why it wanted me to take over immediately, because it should be programed to know to go around a UPS truck.

BERNAL: As a human, you would kind of go over a lane line just smooth and make it a better driving experience. The Tesla doesn’t do that to a dime. And it’s one of those things that sometimes you don’t have to disengage, because it figures it out in milliseconds. But sometimes it can be very jerky because it’s maybe taking a second or two to decide what to do.

However … it’s amazing how in one year they brought a software that was just barely functional, but just like half there, to like almost there now. Just they need to figure out some edge cases. When you’re on the freeways, I trust the system with my life. When you’re in the suburb areas, I’d say we’re 90% there. It’s pretty much solved. However, when we get to the downtown city areas, that’s where it had some very interesting situations.

BERNAL: I had it go down the railroad tracks. I’ve had it almost go into some pedestrians who were jaywalking. And I unfortunately also had a collision, a very minor collision with a plastic bollard in San Jose, where I was making a right hand turn and the car overcorrected into a bollard protecting the bike lane. But luckily there was no people there, and there was no damage to the vehicle or to the city property.

This technology is very advanced, it’s saving people’s lives in real time. It’s even saved my life a few times in real life. It was maybe like midnight. I am driving in San Jose, trying to get some In and Out. I was hungry, you know? And it’s a left-hand turn lane. There’s two turn lanes going left and I’m in the inner left one.

The light turns yellow and I have a Tesla, so I’m not going to lie, I can accelerate through the yellows pretty easily. So I do just that. But the car that was on the outer turn was a little bit ahead of me and they didn’t think I would make the light. So they blatantly just cut into my lane and cut me off, not thinking I was going to run it.

And the Tesla subconscious autopilot kicked in, slammed the brakes before I could even take my foot off the gas pedal.

The average person would have no idea I was on full self-driving in most of my videos. But I do not believe this is anywhere close to safe for an average consumer to get. It takes a lot of mediation and concentration to make the software function appropriately, and that is why it’s beta. And that’s why they give it to select people. So when it is ready, the public can get it.

CHAKRABARTI: That’s former Tesla employee John Bernal. He stress tests Tesla’s full self-driving beta software. But again, it’s not fully autonomous. It’s just kind of called that. But he does that on his YouTube channel, AI Addict.

Now, if you want to hear more about this topic, about how far autonomous vehicle technology has come and what’s yet to be done both programmatically, legally and ethically, do check out our hour-long broadcast today.

This article was originally published on WBUR.org.

Copyright 2022 NPR. To see more, visit https://www.npr.org.