Skip to content
PodSubs
ShowsPeopleGuestsJobsNewsEvents
More
BlogPodCred
Menu
ShowsPeopleGuestsJobsNewsEventsBlogPodCredNetworksSign in
Join nowSign in
  • Shows
  • People
  • Jobs
  • News
  • More
    • Guests
    • Events
    • Blog
    • PodCred
    • Networks
    Join nowSign in

C B m E & U

C B m E & U artwork

Choosing Who Lives with Edmond Awad

Apr 14, 202545 minAudio

Send us Fan Mail Is ethics really just mob rule? Could ethics really be solved with asking millions of people what they think is right? Should an automated car get to decide who lives and who dies? In this episode, Dr. Edmond Awad from the Uehiro Oxford Institute and the Wellcome Centre for Ethics and Humanities at Oxford University to talk about morals and machines, moral machines, and what people's preferences really are about the trolley problem. Remember people, be savvy!

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Transcript

From the show. Tap a time to play from there.

SPEAKER_01The center presents a new podcast, C BME and You, a podcast devoted to promoting awareness and debate around controversial ethical issues in biomedicine.

SPEAKER_04Welcome to episode six, season two of CB Me and You. Today we discuss the moral machine. Is ethics really just mob rule? Can we solve ethics by just asking millions of people what they think is right? And should an automated car get to decide who lives and who dies? Today, we welcome Dr. Edmund Award from the You Hero Oxford Institute and the Welcome with two L's Center for Ethics and Humanities at Oxford University to talk about morals and machines and moral machines.

SPEAKER_04Welcome, Edmund.

SPEAKER_00Thank you for having me.

SPEAKER_02Okay, so our very first question to get to know you a little bit better copy or coffee?

SPEAKER_00Yeah, um I I did try, I think what I thought is copy. Uh it was it was the from uh Yakun Kaya Toast? Yakun Kaya Toast. Um it was just given to me by default. Um and I think it tastes, you know, it it tastes sweet. And I I like sweet. Um I'm not sure if I will be wanting to have a sweet every my everyday drink, but but I yeah, I liked it as a like a delicacy, something that I would have every now and then.

SPEAKER_02And would you prefer it over like a flat white or an espresso or filter coffee?

SPEAKER_00Uh yes, I would prefer it over these things. Really? Um I mean I would probably prefer like a black coffee over it for every day.

SPEAKER_02We'll have to get you to try a um copyo. So it's coffee without any milk.

SPEAKER_04Oh, okay. Yeah. It's not very sweet though.

SPEAKER_02Well, because the beans are roasted with sugar and um butter, and then they will they might still put more sugar in it unless you ask for it sudai.

SPEAKER_04The Sudai without Sudai is less sugar. Did he say koisson? Yeah. It sounded to me like you said koissant.

SPEAKER_02So for listeners, for listeners, um, I bet James that he couldn't sneak the word croissant in today's episode, or I would have to buy munch, and we didn't even make it three minutes before it left.

SPEAKER_04Yeah, too easy.

SPEAKER_02Anyway, moving on. Which country would you like to travel to next?

SPEAKER_00Um, I think Japan.

SPEAKER_02You've got to be like number four or number five. Why Japan?

SPEAKER_00I don't know. I mean, I I like five years ago, I was I I mean I had a trip that was almost, you know, like I almost made it to Japan. Uh and I had a like a presentation to give there. It was all paid for, all prepared, and um everything was going in the right direction. And then um this was supposed to be like end of February 2020. Um and then and then I at the beginning of February, I remember we received like an email saying that the travel, like we had like some kind of risk assessment, saying like this travel assessment to, you know, yeah, is now expanded to seven seven countries, and you have to discuss with your head of department about this.

SPEAKER_00And at the time, of course, you know, it was for me like it's not gonna be a problem. But then um, yeah, my head of department was like, I don't want you to go and then maybe get stuck there. You know, you're teaching now, so it's gonna be a problem. And I thought that was completely unacceptable. Like, how dare she could say that to me? Um, but uh yeah, and I and I had to like write a very apologetic email like to these very nice people who have done a lot of preparation for me to to, you know. Um yeah, so I had to cancel that. So it's still in my heart that I still want to make it.

SPEAKER_04I think after everything that went down, they kind of got it. Yeah.

SPEAKER_02I think you might have been wrong on this one.

SPEAKER_00Yeah, later I felt like at least I don't have to worry about that part.

SPEAKER_02So it sounds like Japan is your long lost love.

SPEAKER_00Yes, do something. Starcross lovers.

SPEAKER_02Excellent. Um last question. Uh what song would you sing at karaoke?

SPEAKER_00Um I think my favorite uh karaoke song is Sway. Uh Michael Bubler uh version. Uh and I think it's it's just because the it doesn't require you to go like a very high pitch in your in your song. So and I have like a voice that is closer to you know, it's it's mostly bariton, but maybe closer to bass uh or bass. I don't know. I don't know. I mean, I don't want to pretend. Uh you know, we'll see uh how it's how my voice is actually being picked up on this mic.

SPEAKER_00Um it's a mic problem, not my voice.

SPEAKER_03With the playback, you're gonna sound like well dance with me.

SPEAKER_02Oh, is that that song? I think so. Yes, yes, yes, it is. So I'm I I'm in the same boat as you. I prefer karaoke songs that are at a lower range, mainly because uh James can sing higher than me. And so the point of being having a feminine voice in the karaoke room is completely mitigated when James is dead. It's like why have I even got it?

SPEAKER_04Yeah, so you're you're a tenor. Well, I'm I'm actually like a bass or baritone. Uh it's just when I I I can I I got I got a pretty decent falsetto, I think. I like to think.

SPEAKER_02Yeah.

SPEAKER_00No, this was gonna be picked up as like, yes, my voice is.

SPEAKER_04But you're a tenor. Oh no. I have a low deep voice. All right. Now I think we need to get to some serious questions and uh make our money. Uh okay, so you're you're pretty uh well known for this thing called the moral machine. Um and obviously this sounds like a technical phrase, and I think a lot of people have might not have heard about the moral machine. Sorry. Uh so can you tell us a little bit more about the moral machine, like what it is, uh what it's used for, and what we can learn from it.

SPEAKER_00Yeah, uh, thank you for um asking me about it. And I think I don't know how famous it is. Uh I think Moral Machine is is a website that is you can find it on moralmachine.net. Uh that the website generates moral random moral dilemmas uh that are basically modeled after the trolley problem. Um and in those moral dilemmas, uh moral dilemmas that are faced by driverless cars. So, you know, it asks you, shows you scenarios that are binary choice, asks you what do you think the car should do in this uh what do you think the uh automated vehicles do or self-driving cars do in these situations.

SPEAKER_00And basically the car is, you know, the brake fails and the car is about to kill one group of people, and then uh the only way out is if it does something and swerves out of its way, uh, which means sacrificing another group of people. And then the website, you know, each scenario will vary different things. Sometimes the car has um some annoying athletes running in front of it. Um uh annoying. Yeah. Very annoying, which you feel like maybe that's what you should do. Uh other times, you know, you might see like, you know, it might be having like a pregnant woman who's like, you know, passing by and you feel like that's probably about more than one person, so maybe it shouldn't do that.

SPEAKER_00Um uh and then basically, so you go like through this like a 13 scenarios, and at the end of it, you get uh some summary of your answers. And this is kind of a gamified aspect of it. So in a way, you could consider it as an online series game or just a just like a as an experiment, a scientific study that is has some kind of gamification elements. Um and um in in basically in generating those random uh dilemmas, um, we've kind of created this as like an uh psychology experiment where you vary several factors like the gender of those uh potential casualties, uh whether they're pedestrians or passengers, whether um they are fit or not fit, uh whether they are like a doctors or or or executives or or like a homeless person.

SPEAKER_00Um so um it's kind of like to s to understand how any of these factors could you know change people's answers.

SPEAKER_04Yeah, great. So I I'm a big enjoyer of trolley problems myself. Uh and I know I know that part of what makes them so appealing, um, at least for philosophers, I don't know about anybody else, is that you can vary all these uh little you can change all these little variables and almost treat it like an like a scientific experiment. You have the same setup, but you change a little variable here, for example, is you say instead of uh letting this person die, you're actively killing them. Um and it can sometimes change the answers that you give.

SPEAKER_04Right? So I'm interested to hear a bit more on like what variables you think are the most interesting. Um when you and when you change them, what do you what makes the biggest impact on what answers people give.

SPEAKER_00Yeah, um we basically varied uh nine different factors. Um and um one of those I think maybe I'll I'll just walk you through those nine factors, and um one of them is the uh as I as I mentioned, that whether the car should stay or swerve. So for us, like this is like a capturing something like omission versus commission, um, killing versus letting die. Um another factor is um whether they're pedestrians or passengers, which is also something that was uh part of the debate as well, like whether the cars should have a loyalty to their um passengers.

SPEAKER_00Um but this on the other hand, uh you might say, well, pedestrians are here externalities, these are people might probably did not opt into being part of this whole thing, so why should they be uh paying the price? Um and um uh the legality aspect, like you know, we thought that maybe if people were crossing legally or illegally might actually impact people's answer. Um and then um the other factors were more related to some features about the potential casualties, like as I said, like their gender, their age uh level, um the fitness level, um social status, uh or you know, we call it social value, I think.

SPEAKER_00Um was it social value? Yeah, something like how much they people think that they contribute to society. Um and um and then the number of of lives as well, which is kind of you know, trying to get into the utilitarianism aspect of it, and also humans versus pets as well. Uh and what we find is actually there are three things that people have kind of approved of more than others, which is the first one is that people prefer or like have a you know approve of the car um uh sparing human lives over pet lives, uh sparing more lives over fewer uh lives, and uh sparing younger lives over uh older lives.

SPEAKER_02I'm really intrigued by the way that you included certain cat uh demographics about people that we might have good reason to not include in automated driving or that what might be actually illegal to consider, such as gender discrimination or age discrimination. Were you were you more looking for a like a discovery to see whether or not people have these preferences? Did you have any worries about what you might find? Like, do you do you think it's justifiable to include those sorts of discriminatory preferences?

SPEAKER_00Yeah, I mean I think it is it is quite controversial to to to um uh to include these things. And I think maybe part of the controversy is also like maybe our uh communication about the purpose of this website was not very clear, and the um uh a lot of the media basically uh picked it up and decided to put their own understanding of it. So uh there was some headlines saying MIT is crowdsourcing the ethics of driverless cars, and in a way, it seemed like this is what's gonna happen.

SPEAKER_00This is how this would be cars will be uh developed with these things. Um of course, from our side, we were uh this was like a psychology study for us, right? Uh so I think this is it's an important thing to do a distinction here. Um and I'm sure your audience probably are familiar with this distinction, which is the descriptive ethics versus normative ethics. Um, and you know, normative ethics is more about what should be done and what ought to be done, uh, while descriptive ethics is more what most people think we should do, or or what our culture tells us to do, or what people behave in daily life.

SPEAKER_00And this is the part that we're trying to capture in the website. So, though, in including those factors, we were more uh, you know, on the side of like, you know, these factors we know from other psychology literature that they have influenced people's decisions about these things. Um so how much do they actually influence the decision in this situation?

SPEAKER_02And how many um how many preferences did you identify? Was it it was in the millions, wasn't it?

SPEAKER_00Yeah, so the you know, to our lucky, you know, happy surprise, uh we had uh we had over millions of participants, um, and we had about 40 million uh dilemmas answered in the first 18 months. Uh now we've crossed the 100 million. Oh, yeah. The website is still around. Thank you. Um yeah, it was, you know, it was I think we did not really anticipate any of that uh uh a priori. Uh we thought, you know, it was, you know, maybe some people will find it interesting, maybe the maybe the nerds and the geeks will probably like talk about it, and and that's it.

SPEAKER_00But we didn't really, yeah.

SPEAKER_04Well, I I put that link in in a course description that I once did. So you're welcome. But the real question I wanted to ask, huh?

SPEAKER_02Yeah, what all four students?

SPEAKER_04Yeah, I don't know how many people actually clicked on it, probably not that many. Um there so you mentioned earlier that uh you know that some people took it and took it very seriously, and some people were um kind of just were more in it for fun. Um did you have a way of differentiating between people who are doing this in the way that you know, as a serious kind of experiment, like the way philosophers treat many trolley problems, or versus people who are just going in to just click about and because they're bored because it's COVID or something like that?

SPEAKER_00Yeah, the short answer is we don't have any way to but but I think I mean I think we think more about it. I guess the the you know the closer comparison is how does it compare to other psychology studies that ask about the trolley problem. Um and in that regard, yes, I mean I think psychologists would prefer to be like more thorough, like try to provide a more uh elaborate description, make sure that sometimes they ask the same questions in many different ways because they know that framing could have an impact. And it is possible that the way our visuals were kind of like uh done in a specific way might have pushed people in one way or another.

SPEAKER_00Uh the fact that the website went viral and made it into some like some of the darkest corners uh of the internet, that also means that we might have got some like a very weird uh uh sample, um, not as an acronym weird, but like just weird. Um and we, you know, and it's true. I mean, we we did have like a you know a substantial, let's say, percentage of people who just wanted to save pets over humans. I don't know if this was like basically uh, you know, people who are serious about it or or people just want to to troll.

SPEAKER_00Like uh we had like maybe like I don't know like maybe 20% or something like uh who had this kind of uh you know, oh save the pets and kill the humans. So yes, in a way we we wouldn't be able to know uh in this case. There were some other issues, of course, like when you ask people what you think the car should do, do you really are you asking them whether they think what they think the right thing is or what they would themselves actually prefer for it to do? Uh so we have all of these issues. Um, and of course, you know, to deal with these issues, you would have to be more like a provide richer way to capture this data.

SPEAKER_00But I think this is the kind of the trade-off right here. So so this is you know, this is what this could capture. There, the reason it could go viral is is because we had to be very people have a very short attention span, so we had to be like a very visual, like a least amount of text, no um, let's say no account registration and all of these things, which would have been nice for us to capture returning users. We had to capture them another way, which is not as reliable reliable. Um, so yeah, it has all of this, you know, of course, it has some methodological uh problems, but uh it also allows us to get into many different countries.

SPEAKER_00So it allows us to do this comparison across countries. Uh and I'm not saying that's also completely out of any problems, right? Um but uh but that's kind of the trade-off. Like if you want to go like a little bit outside and get into outside like a bigger sample, um cross country, outside the Western world, then you have to lose something else. Uh and in this case, we had you had to lose this kind of uh richer way of asking those questions.

SPEAKER_02Did you find um differences in cultures between preferences? Because you said you you found not you looked at nine different demographics altogether, three of which were age saving people over pets and number of people. Number of people. So that that they were fairly common. Were there any nuances between like the East and the West or various regions within each?

SPEAKER_00Yeah, uh so uh that was kind of a very interesting part of the of the paper that we did after is that basically we were able to uh because we had the sheer amount of data from many countries, um, is that we were able to do a lot of interesting stuff, uh like for example, calculating those nine different preferences per each of those countries. And we had like for about, I forgot how many countries, but like maybe over probably 130, like 130 countries, we had like about 100 participants at least from. Um so we for those ones we calculated those nine preferences.

SPEAKER_00Um, and then now we have like a vector of nine dimensions for each of those countries, allows us to calculate distance between countries. Um, and then there's some kind of uh uh clustering technique that allows you to do this hierarchical hierarchical clustering that does not only allow you to shoot to see like three big clusters, which we saw, like three big clusters, but also it could show you for every country what is the closest country as well. So you could see like I should probably have looked up what is the closest to Singapore um before coming here. Uh but basically these three, so in our case, we saw these three big clusters.

SPEAKER_00One of them we called it Western because it has a lot of Western countries, uh, and then one we called it Eastern because it has several countries from uh from Asia and from the Islamic world. Uh and then the third one we call it Southern because it had like a two subclusters that was uh one is Latin American countries and one uh former French colonies. Um and generally speaking, I would say there was some kind of universals and variations. So in terms of in terms of uh the direction of preferences, so let's say if I say, okay, you want to save the younger or you want to save the older?

SPEAKER_00So in most countries you would say people would save the younger, but in some cultures, people like this difference becomes like less pronounced, right? And I think this is more like in the East Asian or you know, you know, uh societies or in Middle East, like you would see that difference goes smaller in terms of uh older. Um other things like uh females over males, that's also something that almost in every country people spared females more males, like I guess female lives is were seen as more precious, I guess, um in some sense.

SPEAKER_00But also there was some kind, you know, that difference was like a shrank in for some different countries. Um one of the most peculiar things that we found is that this um French uh subcluster, and like France specifically, and and the is that they had like a stronger tendency to s to spare the female lives over male lives. Uh and in that in that sense they were slightly different from other European countries, for example, in the Western world. Um, which we you know um in our team we had um we had like a people who are like experts on anthropology and and culture, um who, you know, there was nothing specifically about this in literature at why why women specifically are very highly appreciated in uh in the French culture, specifically, or and other countries that were influenced by France.

SPEAKER_00Um what else? I mean, one of the things that you would expect is seeing that people spread the young over the elderly much higher in the Western world than in the Eastern world. Um there is also we could also do some kind of um basically it's just kind of correlation analysis with other cultural um um indices. Uh one of the things like uh individualism versus collectivism, um, which is you know individualism, uh countries that are individualistic uh focus on the individuals, um, and um they have strong institutions.

SPEAKER_00Um uh people have a loyalty to the state in general, um they have like some kind of pro-sociality to strangers. Um, while in collectivist societies, people care about, you know, they think of themselves as part of a group. Um they you know, family ties are stronger, um, uh institutions are not as strong. Um your loyalty is more to your extended family members than to the states. Um and uh we one of the things we find is that uh in those individualist societies, people were more utilitarian, which is in a way like if you're kind of a you know more like an egalitarian uh way of thinking that people should be, you know, I should treat strangers in the same way I treat my second cousin, then you will think that saving more lives would be more important.

SPEAKER_00Uh, while if you're collectivist, maybe you will think that other factors should be playing in as well.

SPEAKER_02This is absolutely fascinating. Um I'm super intrigued why the French love you know what the French love the women. You're like, um this is really um okay. Um my next question for you is you've garnered so many results from this. You now have um public preferences. So let's just say you took these preferences, you implemented them in the driving in the autonomous vehicles. Um, and now these preferences that the cars have and and an accident happens and it goes the way that people want.

SPEAKER_02And when I say want, I mean last option out of necessity. Does that mean that these choices now are like publicly justifiable? Is there a reason why you really wanted to do this? And and is there a reason why you want to implement these millions of preferences into the way we structure and algorithmitize? Ooh, is that a word? Algorithmize. Help. I'm in a loop.

SPEAKER_04Algorithmatize is what you said?

SPEAKER_02Yeah. Algorith.

unknownHelp.

SPEAKER_00Okay, uh, just first, I mean, I would just want to clarify, we did not program anything into cars based on these preferences.

SPEAKER_02Put that out there. We haven't killed anyone.

SPEAKER_03But we will.

SPEAKER_02We see, we get you, we see.

SPEAKER_04We did you did you wink anyway? Alright, this is a podcast. I need to say wink.

SPEAKER_00Disclaimer.

SPEAKER_05Wink.

SPEAKER_00Uh no, I mean, I guess I guess the fact that it's what the majority, or like say the public, you know, the majority of people think that this is what it should be, does not mean that this is what we should do, right? Um and and I think it's kind of an uh it's kind of an important thing because it also we kind of um it wasn't clear for us like people will see it this way. And so we didn't also work on them on the communication about that part. But throughout the, you know, as we went on and in the project, we started realizing when we get questions about these things, saying people saying, like, well, why can't the car stops?

SPEAKER_00Um like, well, of course the car should stop, right? I mean, if if it can, but like we're talking about if the car cannot stop, right? You know, I think there'd be some people struggled with the with the you know, with the idea of a dilemma itself, right? You know, like when you give them the trolley problem and say, Well, why can't the trolley stop? Um hypothetical.

SPEAKER_03Just go along with it.

SPEAKER_00So I feel like many people, when we started talking about um, so the so in 2016, which is like just uh around the same time where we released the website a few months later, uh uh Germany created this committee of experts uh on um from different aspects of ethics and and legality and all of these things. Uh and those experts sat together and met several times, and um there were like a several disagreements and all of these things, and people did not want to sign off uh because they have like all this disagreement. Um, but you know, they came up with kind of set of things which are I would say you know completely uncontroversial, like uh things like you know, like oh like we should devalue human lives or something like this, right?

SPEAKER_00And one of these things is like, well, the car should definitely try to not sacrifice anyone if it can, if it can. And and I started realizing when I put that sentence on on my slides, people were like, uh, yes, that's finally makes sense. You know, like well, when at what point did you think that um you know, like we're doing this like for the sheer like reason of like, oh, we're gonna we're gonna somebody kind of have to die tonight.

SPEAKER_02Someone doesn't have to, but we will make sure.

SPEAKER_04Somebody's gonna have to die.

SPEAKER_00So uh so I think you know it's it's kind of important to to know that you know, like obviously if the car can do better, it should do better, right? So we're talking about if well moral trade-offs are you know come in a there will be a situation where they're completely uh unavoidable. And in these situations, what what should happen, right? But I guess maybe um in that regard, yes. So so the fact that most people did you know think like, okay, so I think the car should do this or should do that. Now here's what we think about it. So one way you can think of say, oh, like, yes, this is the the people have voted uh to do this, we should do that, right?

SPEAKER_00And this is like a one extreme. And then you have the complete other extreme say, like, well, you know, what do the public know, right? Like, this is like you know, asking a group of people, what do you think this is a square? Do you think this is a circle? Uh and then it's like, oh, the public decided this is so we should just completely like ignore what the public thinks. And I think, you know, like I think both of these opinions are you know extreme, as the extreme says. But then what is really the role of the public in this, right? Um so in a way, like if you tell people like, oh, like, you know, just you know, chill out, like the experts are working on this, right?

SPEAKER_00This is kind of you know, some people might think this is like the best way to do to go over this. And and it many people would probably trust this more, saying, uh, the experts are working on it. But I think it's also important to know what people also think, like just being as being included, right? Like, you know, everyone get a get a say to to express what they think, right? And from a policy perspective, at least, policymakers do actually take the public's opinion in their making of their policies. Um, and I'm not saying that means that they change completely just to cater to the public's interest, right?

SPEAKER_00But they do take it into consideration in different ways. So in some situations you might say, okay, so let's say I have a I'm a policymaker, and I'm just saying, well, the car cannot discriminate based on age, right? And that's something that, you know, backed up by you know ethics and law and all of these things. But now let's say you do this study and you're like, well, a lot of people think the car should actually prioritize the children and the young people because they're vulnerable and you know, like they have more lives ahead of them and all of these things. And apparently in the medical domain, there are situations where you know younger lives are being you know given priority, right?

SPEAKER_00So what happens in that case, maybe in the domain of driverless costs, it doesn't make sense. So in that case, then maybe we should try to you know communicate better why this is the case uh for for the um for the children case, right? Uh now the other case about the number of lives, you might say, well, that's something that not everyone agrees on. Even the German committee who you know who had like all these experts, they said there was a disagreement about having a statement saying that the you know if we're programming the car to minimize the uh the number of lives lost may be justifiable, but the car should not be making trade-offs between lives.

SPEAKER_00So, in a way, like they put like two statements that are not completely uh coherent, and that's also to appease the different groups of people. They had some deontologists uh uh in the committee uh who would not sign off uh some kind of utilitarian uh statement like this. But let's say, I mean, in the case of policy, you have to choose something and work with, right? And in that case, you might say, well, you know, we have a tie here, right? Like we have these two different defensible, ethically defensible um um uh let's say principles, um, but you know, we have to choose one to work with.

SPEAKER_00In that case, you could use the public as a tie-breaking rule, right? Like to choose among the ethically acceptable choices. Um, in the other case, as the children, I say maybe you want to communicate better, maybe in situations where it's about, oh, the public thinks that the higher social status should be prioritized over the lower social status. Well, in this case, you could, you know, safely ignore that after that. Um, so it's kind of like a dealing with disinformation, uh, not completely ignore it, but you know, sometimes you know, taking that into consideration without it being like the having the upper hand.

SPEAKER_02So good law is not mob role. Yeah, that's exactly but good law is still representative law.

SPEAKER_00Yes, to some degree, right? I mean, and and those policymakers are basically hired by a representative who was elected. Yeah.

SPEAKER_04Yeah. So I was I was thinking about uh along the same lines. Um it does seem though that taking into consideration can be a bit of a slippery phrase, right? Because it doesn't specify how you should take something in consideration, either as a tiebreaker or as something that you can use to add on to your reasoning when you already uh agree, or something you should just like look at and be like, oh I've taken this into consideration, moving on. You know, like uh so at least as when it comes to policy, um do you think this is something that can actually make a difference?

SPEAKER_04Um or is it something that you know people might just use to reinforce whatever they they as policymakers happen to already believe in? Um at least unless we really operationalize what taking into consideration means.

SPEAKER_00Yeah, I mean, I don't have any experience in how policymakers make their their decisions and how they usually take this into consideration. Um but my understanding is that they do uh have their own even um or at least maybe in the US they have like some kind of a uh representative sample that they kind of run drafts by and say, okay, what do what do you like at this sample think? And um or at least in some domain or some uh and and and then they just look at it and then I guess yes. I mean they maybe there is no systematic way of looking at it, and maybe that's for the better, right?

SPEAKER_00Because you know, things are messy and you don't want you know to have them to be forced to to take it in a very automatic uh or maybe this is how you know might be the future could be, right? As well. Uh in a way, I I do see a value in in systemizing and standardizing this process because it means there is more maybe coherence and maybe there is more like a fairness that you know in similar situations people treat it equally. Um but also you know, uh it's it could get, you know, I think like these kind of things could be messy as well, and there are some other factors that are you know, sometimes like unspoken rules that should be or should not be uh used.

SPEAKER_00Uh so yeah, I don't know much about I mean I spoke too much for someone who doesn't know much about it.

SPEAKER_04Yeah, fair enough. Yeah, like I haven't thought about this that much, but at least at face value it seems really, really difficult to find a defensible way of operationalizing um how we should take something like the the opinions of the mob into account.

SPEAKER_02Well, I mean that's the principle of procedural fairness. Yeah. So, you know, when decision makers, whether that be policymakers all the way down to the person who decides whether or not you get a license to drive, um, you know, in order for their decision to be procedurally fair, they do have to take into account all the reasons for which they should make or should not make a decision, including your own preferences. And so if you can make the claim that that decision maker did not take into account or did not take into consideration your preferences, which are a good reason for making a specific decision, then you can make an appeal.

SPEAKER_02So we do we do do this. Um it's just, you know, on a grand scale of things, where at what point that goes from good procedural fairness to just mob rule.

SPEAKER_04It's still pretty slippery though, isn't it? Because like if the if the person that you're trying to like take to court can show, oh no, I definitely looked at your the opinion and I thought about it, I just didn't think it was good. Uh then it's like, you know.

SPEAKER_02Well, see that shows that you've taken into into consideration. If you can show that you have reasons for which their preferences or reasons should not be listened to, well then you have proof that you took it into consideration. Take into consideration doesn't mean copy pasting. Yeah.

SPEAKER_00Yeah. I mean, I guess, yeah. I mean, I guess if you if you can you know, if you get this and oh people a lot of people say this, what do you you know, why didn't you take that into consideration? Then maybe if you have already looked into it looked into it, then maybe you already have already answered to it, right? If you have thought about it, I guess.

SPEAKER_02Reason based thinking. I can introduce some Derek Profit.

SPEAKER_04Yeah, but it's not just about drivers driverless cars, because as you mentioned just now, uh many of the lessons that we learn from the moral machine, or at least the the the insights that the moral machine uh makes, is that uh there are certain tendencies or certain uh moral preferences that apply to a whole bunch of things like in healthcare. Um so it might be a chance not just to think about what we should do with new technologies, but also a chance to look at existing policies on, for example, organ donation and take a little take a look at them and reconsider uh uh whether you know um the young should get pol uh organs over the old, or whether we should keep the policies of in of of making sure everybody has an equal chance of getting the organs and and many other examples like that.

SPEAKER_04We don't have to wait for new technologies uh to be developed uh see the applications of um this experiment. Right, right. We can we can look at things that are happening right now in policies that we have now and right, which might have also spread through education as well, right?

SPEAKER_00Like uh you know, this you know, the idea of uh assisted dying.

SPEAKER_02Um I mean that's a bit kind of what about in like um like let's just say there's uh COVID 2.0, there's a new pandemic, then the rate at which people are you know coming into hospital is just so high that we now have an AI technology that can prioritize and triage patients. Um and you know, traditionally this there are some policies, like you know, prioritize call it quality life years, which can be both age and ability discriminatory. But at the same time, it seems to be what people prefer, which is save the most number of lives.

SPEAKER_02So it it seems like the the work you've done could already be applicable in many of these many of these really complex medical decisions that have been being made for a really long time.

SPEAKER_00We did actually um just capitalizing on the moral machine um uh popularity, uh, which you know we still have people coming to the website. So during COVID, we also said, okay, we want to see this ventilator dilemma. And that was um uh that was the time before the vaccines were developed. Uh uh so we did ask it was it was not very gamified, it was more like uh this you know very boring kind of style of asking questions about what do you think? Um how do you think, you know, like how how like do you think the vaccine should be very prioritized?

SPEAKER_00Like and they say, okay, do you think it should prioritize should be given like you know, younger people should be given more priority? Or you think people who are have a um higher chance of surviving from this, right? Or or you know, quality life and and and these things. And one thing we found was kind of and we also included two things, which is like uh random allocation, like uh and and the first conference served. And those were like more, I would get say like kind of deontological kind of version of it.

SPEAKER_00Uh and interestingly, we've I mean we did this in like so we did this in like a we had like a 20 countries, but we also did uh uh national presentative samples in four different countries in Brazil, Japan, USA, and um maybe Korea. Uh and we found is basically there was some kind of sort of like polarization. So we found that in all of those countries, the most two popular uh options is that one of them would say like yes, use some factors to prioritize. Like yes, the younger should have more, yes, the one who's gonna survive, have more, yes, the one who um um I don't know if there was a gender one, female over male, I just have to remember.

SPEAKER_00Um but yeah, quality of life, it should be prioritized, and then you have a complete opposite group which said that like these two of you know random or of best first come, first serve should be the one that is the most. So out of all the possible combinations, because there could be a lot of different combinations of this by people saying, like, okay, only if it's like a prognosis situation, you can like no way. So we have like a this kind of real polarization, and then our next question saying, Okay, so you know, at the country level, even at at any country's level, you're gonna have a problem if you're gonna put you know something like that.

SPEAKER_00And so then we ask people like basically we try to look in our data, like what is the thing that is least hated in these things? And it seems like prognosis was the thing that people say, okay, you know, if you're gonna prioritize people based on this likelihood of you know survival from these things, you know, I guess that's kind of okay, right? So uh and so yeah, so some of these other factors, you know, the moral machine uh like factors did play in in this case, but somehow it wasn't just cultural, it was more like a country level.

SPEAKER_04Yeah, wow. And I think that there over COVID at least, there are ways in which you could see people's moral preferences change. Like at the beginning, a lot of people were saying, oh no, we should go for we should try to go for a COVID zero type policy to ensure that as we save as many lives as possible. But one and a half years in, people are like, eh, you know what? It's a risk worth taking. Uh I just want to get out of the house. So I guess my my question is like if you were to run the moral moral machine or something like that again, are there any like uh new variables that you'd want to include in this new experiment?

SPEAKER_00Right, because you're thinking about it in terms of like, okay, contribution or social contribution in in that sense.

SPEAKER_04Yeah, or like just given between COVID and new technology and AI and all that, uh do you think there are any new uh moral considerations that are worth trying to uh fit into an experiment and see what people might think about them?

SPEAKER_00Yeah, I mean I'm thinking that uh one of the things that I was really surprised became like something that is debatable is vaccines, right? Like uh COVID vaccines specifically, right? And this is also like a debate that is politicized in the US, right? And you know, like um at the time, what we knew at the time, right? It seems like you know, this is the best strategy because because of this group, you know, this social effect, right? So um, and people who studied epidemiology understand that that this case of COVID is not a case of car crash, right?

SPEAKER_00Like, you know, car crash do not spread to other car crashes. I mean, they do spread maybe very locally, but uh specifically in that very moment, but they don't really spread around like in the same way that COVID does. So like uh doing a policy, I mean, even the policy before before the vaccine was like you know, limiting people's freedom of going out, which a lot of people were also like completely outraged by. It's like saying, you know, my my you know, my individual freedom now. I'm being forced to stay at home, I'm being forced to wash my hands uh after going to the toilet. I mean, what's this crazy things?

SPEAKER_00Um I'm being forced to uh to go and uh take this vaccine, which has some side effects, you know, and and like and you know, so uh which is kind of this it does relate to like omission versus commission, like in a in a way like oh, if you don't take the vaccine, you know, yeah, maybe nothing will happen to you, but you know, your grandma might not really survive, right? But if you do take it, well, hey, something bad could happen to you and your grandma, you know, like uh so it's kind of and and if you don't take it, you something could happen to you as well, right? So um I think you know, like this is like a some topic that is not easy to understand because you have to un you know, you have to understand epidemiology spread.

SPEAKER_00People would not understand that if 10 people were infected today, it means 100 people could be infected tomorrow, and it means like you know, 1,000 people the day after. So this thing you were you know, we're not cognitive, you know, cognitively we're not really uh we did not evolve to think in this kind of at this kind of level of scale. So it's not immediately think to see it this way. So um I guess you know, to this question, I would probably think more about uh like this kind of how to understand like you know, things that are let's say um um that are simple ideas that you know versus complex, you know, like a you know things that could, you know, complex systems like in the case of uh COVID.

SPEAKER_04Um I I feel like I didn't really communicate this very well, but maybe there's a possible experiment somewhere that you you can ask people. Like imagine you're in a lockdown, but let's say we change one policy at a time. You know, you you you can now go to the bars instead of uh not going to the bars. Uh but if we do that, you know, on average maybe an extra thousand people die. Yeah. Uh what what do you think should be done here?

SPEAKER_00Well, something like that. I like that. And I think this also has this, I think it has also this educational aspect to it as well, because I think people need to see uh like we need more tools that could actually allow people to like you know play and see how how things change, right? Like, you know, like some sort of like a simulation saying, like, okay, here are the policies, you can decide which policy should be done in in COVID, and then it gives you the number of people who will die out of this, you know, out of each of these policies. Like, you know, yes, you're gonna restrict more people, but this is how much difference it would make, right?

SPEAKER_00And when people realize that, oh, actually, for some reason it is making a big difference, right?

SPEAKER_02Um, yeah, I I find that really fascinating. I think because everyone's so willing to um judge and to say what they think would be the right case, and then you hand them the consequences of their decision, particularly at the universal scale, and maybe they might not be so quick to judge next time.

SPEAKER_05Yeah, yeah.

SPEAKER_02That's all the time we have for today. Thank you so much for coming on the pod. Uh, do you have any exciting news or any more websites coming out that James and I can waste an afternoon gamifying moral philosophy?

SPEAKER_00Thank you for asking me this question. Uh, we actually do have a website um that is models like a similar style of the moral machine. It's called Tinkertots. Uh, and it's basically about the embryo selection and based on polygenic screening. Uh so the idea is that you know, you imagine that you know you're a couple who are having some kind of uh IVF, you have several embryos, and you need to choose which ones you want to implant. And now some companies offer, well, companies initially offered doing some kind of testing for like genetic testing for some of these uh conditions.

SPEAKER_00Uh, but now it seems like the selection has you know moved into polygenic screening, which includes several things that is not just medical conditions, but also some non-medical uh traits. Um and you know, it hasn't expanded too much so far. I mean, these are mostly in the US as well. Uh, but in this case, we kind of try to explore how much it would impact people's decisions to know that okay, you have this embryo that could have uh above average chances of having a low IQ, uh, this one has a above average chances of being of having an antisocial behavior. Uh, and like which one would you choose uh to to implant?

SPEAKER_00Uh so you would, you know, if you go on uh tinker-tots uh.net, uh, then you can see the websites should be going live in an hour from now, actually.

SPEAKER_02Oh, perfect.

SPEAKER_04I'm so excited.

SPEAKER_02Yeah, yeah. Looks like we'll be able to do that this afternoon.

SPEAKER_04Uh yeah, I'll be doing my real job.

SPEAKER_02This is our job. Um, all right, we'll add the link to the podcast um description. So thank you for that. Congratulations on a on a new website, and hopefully you get a hundred million more preferences.

SPEAKER_00Thank you very much. Thank you.

SPEAKER_02We'll take you out for some coffee. That's busy. All right, thank you.

SPEAKER_01Thanks for listening to our new podcast. Subscribe online and tune in every first of the month. In the meantime, listen to CBME and you and be savvy.

Open the transcript file

Hosts and Guests

Nobody is credited on this episode yet.

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. Can Nihilism Improve Your Mental Wellbeing?

    Nov 2, 202645 minAudio

    Send us Fan Mail We spend much of our lives searching for meaning. Some of us look for it at work. Some in our families and friends. Some look for it at an overpriced spiritual retreat in Bali where some dude named...

    Credit the people on this episode

  2. Insurance Discrimination: Is it Fair? with Harisan Nasir

    Sep 24, 202649 minAudio

    Send us Fan Mail In some insurance scheme, you might be required to pay more because you are more prone to disability. In another, you may have to pay more because you're a woman (and are statistically likely to live...

    Credit the people on this episode

  3. Genomic Newborn Screening: Where less really is sometimes more with Ainsley Newson

    Aug 25, 202655 minAudio

    Send us Fan Mail Did you know that genomic sequencing can now screen newborns for hundreds of rare conditions right at birth? That looking too closely at our DNA might actually cause more anxiety than answers? And is...

    Credit the people on this episode

  4. The Hidden Side of Birth: Injury, Trauma, and PTSD with Angela Ballantyne

    Jul 2, 202648 minAudio

    Send us Fan Mail Did you know that over 80% of people who give birth experience an injury? That anywhere between 30 and 50% of people who give birth find it traumatic? And that up to 15% of people experience PTSD from...

    Credit the people on this episode

  5. Brains in Jars with Insoo Hyun

    Jun 1, 202656 minAudio

    Send us Fan Mail Could a blob of cells in a petri dish one day think? Feel? Have rights? In this episode, Professor Insoo Hyun (Harvard Medical School, one of the world's leading experts on the ethics of stem cell and...

    Credit the people on this episode