SDS 148: The Trolley Problem

SDS 148: The Trolley Problem

Trolley ProblemWelcome to episode #148 of the Super Data Science Podcast. Here we go!

Today it's Five Minute Friday time

The Trolley Problem, as Wikipedia puts it, is a classic thought experiment in ethics.

But what if I told you it’s no longer just a theoretical exercise, but a fast approaching reality that society is going to have to grapple with? That this actually could be of determining who lives and who dies in certain situations?

The self-driving car doesn't just throw up technological challenges, but questions of morality too. And that’s exactly what we’re going to explore on this latest episode of FiveMinuteFriday.


  • How would you answer both scenarios of this Trolley Problem? Does this change your attitude towards self-driving cars?
  • Download The Transcript
  • Music Credit: Limitless by Elektronomia [NCS Release]


Full Podcast Transcript

Expand to view full transcript

This is Five Minute Friday episode number 148, The Trolley Problem. Welcome back to the Super Data Science podcast. Today we're going to have an interesting, and at the same time, controversial discussion. You probably have heard that you need to surround yourself with amazing people, with interesting people that you want to be like, learn from, and therefore you will grow. I try to do that in my own life, and one of these amazing people in my life introduced me to something that is very exciting, and it's called the Radiolab podcast. It's a podcast which you can find on iTunes, and I'm assuming that since you're listening to this podcast, you are interested in podcasts, so definitely check it out.

Radiolab is a high-production show that does research into current topics, or any kind of different topics that are interesting, that are exciting, and they produce them with very high quality, great audio, great audio effects, lots of guests, lots of different comments from different people that are cut into the episodes. The research is very deep, so so far I've listened to two episodes. One was about gun control and that one was an hour long, and they went through the whole history of the second amendment, how it only actually started being interpreted as it is now in the 2000s. This whole interpretation actually started in 2008, as I believe, after a case in Washington.

Before that, there was something happening in 2001, and then they went to the history from the 1960s, so it was a really cool episode.
Today, I listened to an episode called ... What was it called? Let me quickly have a look, Driverless Dilemma. The episode was called Driverless Dilemma. What they were talking about is what I wanted to share today. I'm only going to share a snippet of what I learned. If you want to learn the full story and all the research, like they talk to an MRI scientist and research the human brain on this subject, then check out Radiolab. But let's get started. We're going to talk about the driverless dilemma. Actually, we're going to talk about the trolley problem, which is part of the driverless dilemma.

The trolley problem goes like this. You are standing on a trolley, which is going down some train tracks. Along those train tracks, ahead of you, there are five workers working on the track. They're facing away from the trolley, they cannot see you coming, and they cannot hear the trolley, and you cannot yell out to them. If you do nothing while on this trolley, then what'll happen is that all five workers will die. On the other hand, it so happens that on the trolley near you, there's a lever, and if you pull that lever, then the trolley will divert onto some side tracks where there's only one worker, and that worker will die.

In this case, your choice is A) do nothing and five people will die, or B) pull the lever and one person will die. Question is, what will you choose? Think about it for a second and just have your answer with you. I really wonder what you selected, because when the guys from Radiolab asked this question from, I think it was random people on the street, 90% of the people said they would pull the lever. They would kill one person instead of killing five. They basically would pull the lever, kill one, instead of do nothing and kill five.

Then, we go to part two. Now imagine you're standing above the train tracks. You're standing on a bridge that goes over the train tracks, and you can see the trolley approaching, and it's going to, as soon as it passes the bridge, on the other side, it will kill ... passes under the bridge, it will kill those five same workers. The only thing you can do now is, again, the two things you can do, you can do nothing,
version A like last time, and it will kill the five workers, or version B, there's a ... You notice that there's a large man standing near you, a large person standing near you on the bridge, and what you can do is you can push the large person off the bridge onto the tracks. He will die, but he will stop the trolley and therefore save the other five people. The question is will you do A, nothing, and let five people die, or will you do B, push the large person off the bridge onto the tracks, and thereby stop the trolley and kill that one person, but save the five?

What's your instinctive response here? Let me guess, probably you're going to say no, you're not going to push the person. Why can I guess that? It's because when the guys from Radiolab did the same survey, but with this question, most people, 90% of the people said that they wouldn't kill the person, because in this case it really feels like murder. It feels like you're murdering a person, even though you're trying to save five. Very controversial topic, very controversial question. Good thing that it's just theoretical. It's just theoretical, we're never actually faced with a choice like that in life. This is something that we'll just contemplate about in the trolley problem.

But the thing is, why it's now become current, is because of the proliferation of self-driving cars. Self-driving cars are starting to pop up. There are already self-driving Ubers which you can take. By 2021, certain companies in Germany are going to release self-driving cars for sale to the public, so you can actually ... You'll be able to own self-driving cars, might even happen sooner with other automobile manufacturers or companies in the US. This is the future we're going into. Self-driving cars are going to have to have pre-programmed algorithms that will allow them to make choices like that.

For instance, if there's people on the road, pedestrians on the road, if there's five pedestrians on the road. Let's think about this theoretically. Again, they gave this example in the Radiolab episode. There's five pedestrians on the road, and there's a self-driving car coming towards them. Should the car kill the five pedestrians, or should it run into a concrete wall, and thereby kill the passenger of the car? Those are the only two options it has. The circumstances are such that there's nothing else it can do, and it can only decide between the two. How does it decide? What is the correct decision? Most people answer to that question that it should sacrifice the person inside the car. It should sacrifice the passenger in order to save five people. Sacrifice one life in order to save five. But then, when those same people are asked will you buy a car like that? Will you buy a car that is pre-programmed to intentionally kill you in order to save more lives than just one, most people said no. Most people said they won't buy a car like that.

Now, we have a question of morality, and also how are these companies going to be perceived by the public? Are people going to actually buy the cars? Are they going to be perceived as moral or amoral, and so on? All of this ties in very intricately into the question of ethics in data science, machine learning, and artificial intelligence, because these questions, they need to somehow be addressed in advance. They cannot be left up to some programmers who are creating these algorithms to decide this on the spot. These things have to be thought through not just at a company level, they should be thought at a national level, or even a global level.

For instance, Germany's one of the first countries that has passed a law, as I understood from the podcast, that they've passed a law that addresses the issue of discrimination, that autonomous vehicles should not discriminate between people on any basis. They should not discriminate based on gender, on race, on age, on social status, on income, or on anything like that. Yes, that is indeed a possibility that, for instance, self-driving cars could discriminate potentially even on things like income level of a person, because eventually they'll be able to communicate with each other. They'll have so much data about us that they will know who's sitting in Car A, who's sitting in Car B, and potentially they could make these decisions based on who has a more affluence status in the socium, or who is younger, or who has a ... Maybe somebody has a terminal illness, and those might not be ethical things. While there's no right or wrong answer right now, that we can think of, these are things to keep in mind and to consider, and this is where the world is going. Self-driving cars, they're coming into our lives very rapidly, and these are questions that we will need to be addressing.

There's something to ponder on, something philosophical to think about. Maybe a topic starter for you for this weekend, if you're going to be chatting to some friends, or going to an event, social gathering, or something like that. See how people react to the trolley problem, or the driverless dilemma. Of course, if you're interested to learn more, highly recommend the Radiolab podcast. This was Driverless Dilemma, which was released on 27 September, 2017. The episode is 42 minutes long. Highly recommend checking it out if you're interested to learn more, and understand not only philosophically what's happening, and what the dilemma is, but actually from a neurological perspective, what happens in the brain, because they talk to a scientist who's put people into an MRI machine and studied their brain when they were answering these questions. Plus, there's some more interesting questions that are discussed in the show.
All right. Hope you enjoyed this short excurse into the world of philosophical debates and autonomous vehicles that are going to be more and more current in years to come, and I look forward to seeing you back here next time. Until then, happy analyzing.

Kirill Eremenko
Kirill Eremenko

I’m a Data Scientist and Entrepreneur. I also teach Data Science Online and host the SDS podcast where I interview some of the most inspiring Data Scientists from all around the world. I am passionate about bringing Data Science and Analytics to the world!

What are you waiting for?


as seen on: