The ethics of self-driving cars
20:19 Share

The Moral and Ethical Challenges of Self-Driving Cars Hi there, everyone. It's Jeff, and this is Plain English on Monday, March 24th, 2025. Thanks for starting your week with us and for including Plain English in your learning routine.

If you're new to Plain English, here's how it works. Every Monday and Thursday, we talk about something going on in the world, something in the news. And you get to learn new words and new ways of expressing ideas. But best of all, you get to start thinking in English. And this is another episode that will make you think.

It's about the ethics behind self-driving cars. What I mean by that is, engineers who design the software have to make some hard life and death decisions about what the cars should do in dangerous situations. And these are often uncomfortable questions.

Now, just a quick warning. We are going to be talking about life

death, injuries, and danger. Nothing too gruesome, but if that's not your vibe today, you can listen to a different episode about cars. I recommend you go back and find episode 435. That's when I drove a Tesla and reported back what it was like.

But it's time to get started on this episode. This is number 760. So just as a reminder, you can find the full lesson resources at plainenglish.com slash 760. Before we start today's story, I'd just like to remind you that the podcast is just one part of how we can help you upgrade your English skills.

At plainenglish.com, you can make faster progress with active learning strategies. You can take quizzes, do activities, listen to the fast version of the audio, watch video workshops, practice what you learn, and even join a live call with JR and me. It's all about helping you build your skills to become a better, more confident English speaker.

Sound good? Go to plainenglish.com to start your free 14-day trial today. Now, let's jump into today's story. The trolley problem is a common thought experiment in philosophy. Imagine you're out for a walk and you see a runaway trolley. A trolley is like a one-car train.

The trolley is out of control and can't stop. Five people are tied to the trolley tracks. If you do nothing, the trolley will strike and kill the five people. But just ahead, there's a switch with a handle. If you pull the switch, you'll divert the trolley onto a different track.

If the trolley goes on that other track, just one person would be killed. What should you do? Should you do nothing and let five people die? Or should you pull the handle and send the trolley in a new direction, killing one person who would otherwise live?

Like many questions in philosophy, the trolley problem doesn't have a clear right or wrong answer. Your answer reveals your values and how you think about a problem. If you do nothing, you'll be a witness to a terrible tragedy. Five people will die in an accident that you did nothing to cause.

If you pull the switch, just one person will die. Well, that's better than five people dying, probably. But you will have been an active participant in a person's death, not a bystander. Luckily, nobody encounters the trolley problem directly in real life. That's what makes it such an interesting thought experiment.

But situations like this do arise in everyday life and nowhere more than on the road. Like it or not, roadways are dangerous, and every day human drivers make decisions that affect not just their own lives, but also those of others.

Do you brake or swerve? Do you hit an animal or a parked car? Do you swerve into a tree or into the other lane of traffic? To make these decisions in a split second, human drivers draw on their experience, morality, and intuition, flawed though it may be.

But self-driving cars have no morality or intuition. Self-driving cars have never pondered the trolley problem. But these advanced machines will have to make lightning-fast decisions about life and death. So how are they going to do that?

There are two pieces to the puzzle, math and morality. Start with the math. The software in autonomous vehicles assesses the probability of different outcomes. For example, if a pedestrian is standing on the sidewalk, there might be a 10% chance that she steps out into the roadway.

If she takes one step forward, that probability would increase to nearly 100%. So the car is constantly assessing the environment. And it can take certain actions in response to what happens in the outside world.

And every action the car takes has a potential cost and a potential benefit. If a car breaks suddenly, there's a cost to passenger comfort, but probably a benefit to safety.

If a car swerves out of the way of a pedestrian, there's a benefit to the pedestrian but a potential cost to the drivers in adjacent lanes. All this data can be combined to determine a range of outcomes, each with a cost and a benefit.

That's the math. But the car and the car's engineers cannot make the right decision without a sense of right and wrong. And that's where the morality comes in. Most people think self-driving cars should be utilitarian. They should maximize overall safety and minimize overall danger.

Given the choice between putting one person at risk and five people at risk, most people agree that the car should put one person at risk. But real-world situations are often not that simple. Here's an example.

Imagine you're in the middle lane of a three-lane highway. Okay? You're driving in the middle lane. To your left is a car with five passengers. To your right is a motorcycle. Directly in front of you is a truck.

So you've got a car to the left, a motorcycle to the right, and a truck in front of you. Without warning, cargo starts falling off the back of the truck. You or your car's computer have several options. You can hit the brakes. If you do that, you risk being hit by the car behind you.

You can do nothing and hit the cargo, but who knows what will happen then. You can swerve to the left, but remember, what's to your left? It's a car with five people in it. That's not good. You can also swerve to the right. If you do that, you hit a motorcycle. Now, if you hit the car, you put five people in a little danger.

If you hit the motorcycle, you put just one person in danger, but it's a lot of danger. If you slam on the brakes, you endanger the person behind you, and you also endanger yourself. So what should you do?

Well, don't worry, you don't have to decide now. If you are ever in that situation, your body will react before you even have time to think about it. But programmers of self-driving cars do have to weigh decisions like that as they program the car's software. Is it better to put more people in a little danger

or one person in a lot of danger. Which is better, a 50% chance of a two-car crash or a 5% chance of a 10-car pileup? Which should we choose, $5,000 in damage to a car or the life of an animal on the street?

One of the biggest ethical dilemmas is the relative value of the person in the car versus a person out of the car. Imagine a situation where a pedestrian steps in front of a vehicle. The car can either swerve and hit a pole and kill or injure the person inside,

Or, the car can stay the course and hit the pedestrian. What should it do? A researcher posed this question to study participants, but with a twist. He asked people whether they, as human drivers, would swerve and put themselves in danger...

or if they, as human drivers, would stay the course and hit the pedestrian. Then he asked a related question. He asked what a self-driving car should do in exactly the same situation, and he compared the responses. The results might surprise you. Respondents said that if they were driving...

They would swerve and put their own lives in danger. So they would act against their own personal interests if they were driving. They would save the other person. But the same survey respondents said if they were riders in an autonomous vehicle, the vehicle should stay the course and hit the pedestrian.

The autonomous vehicles should protect the passengers. Now, why do people think that? People want the self-driving car to protect the passengers, but the human-driven car to protect the pedestrian. Why?

One theory is that when we are in control of a vehicle, we feel a moral obligation not to harm others, even if it means putting our own safety at risk.

But when we are passengers in autonomous vehicles, we have no control, and therefore we selfishly want the machine to save us and not the other people. These are hard questions, and there is no objectively right answer to them, just as there is no right answer to the trolley problem.

But it does mean that the makers of self-driving cars have to essentially build a moral code into their products. And informed consumers are concerned about ethical questions like the ones we've talked about today. A recent survey asked 1,600 people about the challenges to adopting autonomous vehicles.

There were technical, legal, and ethical challenges. And respondents said the question of saving the pedestrian or saving the passenger was the biggest challenge of all, higher even than technical challenges like driving in bad weather. So makers of self-driving cars should not ignore ethical questions like these.

They'll have to have good answers if they want to convince the public to buy a self-driving car or pay for a ride in one.

Don't let these hard questions put you off self-driving cars. I've said before, I love driving. I love the feeling. I like being behind the wheel. But a world of self-driving cars is just better for everyone if the technology can be made to work. So just think, no more drunk driving. No more distracted driving.

No more falling asleep behind the wheel. The elderly and disabled can have the mobility that other car owners enjoy. And the word car owner may be a thing of the past. If there are cheap self-driving taxis, no more car loans, down payments, interest, repossessions.

All the time you spend concentrating on the road can be freed up to do other things, but

The waste of paying for insurance, repairs, and hospital bills that result from human error, all that can be significantly reduced. But there is a long way before we get there. And as we've learned, it's not just the technical issues, but also these moral and ethical issues that have to be resolved.

All right, that does it for us here at Plain English. Congratulations on making it to the end of another episode. And remember, you can find the full episode at plainenglish.com slash 760. We'll be back on Thursday with a new topic. See you then.

Listen up if you speak Spanish, Portuguese, Chinese, German, French, Italian, Japanese, Polish, or Turkish. One of the most frustrating things about listening and reading in a second language is being confused by unfamiliar words or phrases.

When you listen to plain English, or when you read the transcript, you probably encounter some words you don't know. When that happens, you have a few options. You can stop the audio, go look up the definition, then return to the episode, find your place, and press play again, but that's exhausting.

Or you can just skip it and be confused. That's no fun either. But what if you could see the translation of the word into your language instantly without having to stop the audio, without having to look anything up?

Well, you can at plainenglish.com. For each episode, we translate about 100 words and phrases from English to nine languages, Spanish, Portuguese, Chinese, German, French, Italian, Japanese, Polish, and Turkish. And we highlight those words in blue.

All you have to do is hover your mouse over the highlighted word and you'll see the translation instantly. It works great on mobile too. It really makes listening a lot more satisfying and it's a great way to expand your vocabulary in English.

The best way to see these translations is to sign up for a free 14-day trial at plainenglish.com. During your trial, you can read as many transcripts as you like, all with the translations into your language. So sign up for your free trial today at plainenglish.com.