Making Sense - Sam Harris - #56 — Abusing Dolores

00:00:00.000 Welcome to the Making Sense Podcast.

00:00:08.820 This is Sam Harris.

00:00:10.880 Just a note to say that if you're hearing this, you are not currently on our subscriber

00:00:14.680 feed and will only be hearing the first part of this conversation.

00:00:18.440 In order to access full episodes of the Making Sense Podcast, you'll need to subscribe at

00:00:22.720 samharris.org.

00:00:24.140 There you'll find our private RSS feed to add to your favorite podcatcher, along with

00:00:28.360 other subscriber-only content.

00:00:30.260 We don't run ads on the podcast, and therefore it's made possible entirely through the support

00:00:34.640 of our subscribers.

00:00:35.880 So if you enjoy what we're doing here, please consider becoming one.

00:00:46.680 I have Paul Bloom on the line.

00:00:48.800 Paul, thanks for coming back on the podcast.

00:00:50.640 Sam, thanks for having me back.

00:00:51.860 You are now officially my, well, I have, I think, only two return guests.

00:00:57.620 But you have just edged out David Deutsch, who has two appearances.

00:01:02.080 So you're the only third appearance on this podcast.

00:01:05.040 So that says something.

00:01:06.500 It's not exactly like a 20th appearance on The Tonight Show, but it is a measure of how

00:01:11.080 good a guest you are.

00:01:12.220 I'm touched.

00:01:13.000 Maybe, you know, maybe a decade from now.

00:01:14.480 Who knows?

00:01:14.860 We could be doing our 20th anniversary show.

00:01:16.760 Well, after we did our second show, people just emailed me saying, just have Paul on the

00:01:22.020 podcast all the time.

00:01:23.580 You don't need any other guests.

00:01:24.960 So you are a popular guest.

00:01:26.880 Well, we had a great discussion.

00:01:28.300 I think a little bit of what makes for a good discussion, which is you and I agree on a

00:01:32.360 lot.

00:01:32.740 We have a lot of common ground, but there's enough tension and enough things to rub against

00:01:36.920 that we get some good discussion going.

00:01:39.120 We will see if we can steer ourselves in the direction of controversy, perhaps.

00:01:43.320 But you have just released a book, which we talked about to a significant degree, I think,

00:01:50.220 in your first appearance here.

00:01:51.600 And we would be remiss not to talk about it some.

00:01:55.460 So we'll start with that.

00:01:56.660 But people should just know that if they find what we were about to say about empathy intriguing,

00:02:02.760 our first podcast has a full hour or more on it, and it is an incredibly interesting and

00:02:10.020 consequential issue, which we will be giving short shrift here because we've already done

00:02:14.740 it.

00:02:15.100 But the proper intro to this topic is that you have just released a book entitled Against

00:02:19.940 Empathy, which is a, I think I told you at the time, is a fantastic title.

00:02:25.260 You seem to steer yourself out of a full collision with the outrage of your colleagues in your

00:02:30.680 subtitle.

00:02:31.300 You have, as a subtitle, The Case for Rational Compassion.

00:02:35.540 So you're not against compassion, obviously.

00:02:38.060 Tell us about your position on empathy and how it's different from compassion.

00:02:45.220 So the distinction is super important because if you just hear him against empathy, it'd be

00:02:49.580 fair enough to assume I'm some sort of monster, some sort of person arguing for pure selfishness

00:02:54.900 or, you know, entire lack of warmth or caring for others.

00:02:59.060 And that's not what I mean by empathy.

00:03:00.840 And it's actually not what psychologists and philosophers mean by empathy either.

00:03:05.120 What I'm against is putting yourself in other people's shoes, feeling their pain, feeling their

00:03:10.660 suffering.

00:03:11.540 And I'm not even against this in general.

00:03:14.080 I think empathy is a wonderful source of pleasure.

00:03:16.760 It's central to sex.

00:03:18.440 It's central to sports.

00:03:20.020 It's central to the pleasure we get from literature and movies and all sorts of fictional entertainments.

00:03:25.780 But what I argue is in the moral realm, when it comes to being good people, it steers us

00:03:32.220 dangerously astray.

00:03:33.820 It's a moral train wreck.

00:03:35.560 And the reason why is that it zooms us in on individuals like a spotlight.

00:03:39.720 And in fact, the fans of empathy describe it as a spotlight.

00:03:43.720 But because of that, it's very biased.

00:03:47.360 I'll be more empathic towards somebody who is my skin color and then of a different skin

00:03:51.300 color towards somebody I know versus a stranger.

00:03:53.740 It's difficult to be empathic at all to somebody who you view as disgusting or unattractive or

00:03:59.900 dangerous or opposed to you.

00:04:02.760 And in fact, there's a lot of neuroscience studies we can get into that get at this not

00:04:07.080 only through self-report, which is kind of unreliable, but actually looking at the correlates

00:04:11.160 of empathy in the brain.

00:04:12.960 You know, finding that some studies find that one of my favorite studies tested male soccer

00:04:17.460 fans in Europe.

00:04:18.260 And they watch somebody who's been described as a fan of their same team receive electric

00:04:24.460 shocks.

00:04:25.480 And then it turns out they feel empathy.

00:04:27.040 In fact, the same parts of their brain that would be active if they themselves were being

00:04:31.000 shocked light up when they see this other person being shocked.

00:04:34.200 So that's great.

00:04:35.100 But then in another condition, they observe somebody who's described as not being of the

00:04:40.120 supporting the same team.

00:04:41.420 And there, empathy shuts down.

00:04:44.100 And in fact, what you get is kind of a blast of pleasure circuitry when they watch the other

00:04:48.960 person being shocked.

00:04:50.640 And so empathy is biased and narrow and parochial and I think leads us astray in a million ways,

00:04:56.880 much of which we discussed the last time we talked about this.

00:05:00.260 Compassion is a bit different.

00:05:01.380 So my argument is what we should replace empathy with for decision making is cold-blooded reasoning

00:05:08.300 of a more or less utilitarian sort where you judge costs and benefits.

00:05:13.180 You ask yourself, what can I do to make the world a better place?

00:05:15.920 What could I do to increase happiness, to reduce suffering?

00:05:18.920 And maybe you could view that in a utilitarian way.

00:05:21.380 You could do it in terms of a Kantian moral principles way.

00:05:24.600 But however you do it, it's an act of reason.

00:05:28.000 What's missing in that, and that's the rational part of my subtitle,

00:05:30.880 what's missing in that is everybody from David Hume on down has pointed out you need some sort of

00:05:36.160 motivation, some sort of kick in the pants.

00:05:38.420 And that's where I think compassion comes in.

00:05:41.160 So many people blur empathy and compassion together.

00:05:43.800 And I don't actually care how people use the terminology.

00:05:46.520 But what's important is they're really different.

00:05:49.060 So you can feel empathy.

00:05:51.460 I see you suffer and I feel your pain and I zoom in on that.

00:05:55.060 But you could also feel compassion, which is you care for somebody.

00:05:58.320 You love them.

00:05:59.260 You want them to thrive.

00:06:00.320 You want them to be happy, but you don't feel their pain.

00:06:03.720 And some really cool experiments on this, for instance, were done by – and this is going to

00:06:07.440 connect to one of your deep interests, that of meditation – were done by Tanya Singer,

00:06:12.540 who's a German neuroscientist, and Matthew Ricard, who's a Buddhist monk and so-called

00:06:17.120 happiest man alive.

00:06:18.180 And they did these studies where they trained people to feel empathy, to experience the suffering

00:06:24.580 of others.

00:06:25.600 And then they trained another group to feel compassion.

00:06:28.240 And the way they do it is through loving kindness meditation, where you care about others,

00:06:32.280 but you don't feel their pain.

00:06:34.100 Now, it turns out these activate entirely different parts of the brain.

00:06:36.940 There's always some overlap, but there's distinct parts of the brain.

00:06:39.220 But more to the point, they have different effects.

00:06:42.280 So, the empathy training makes people suffer.

00:06:45.240 It makes people selfish.

00:06:46.480 It leads to burnout.

00:06:48.040 While the compassion training is pleasurable.

00:06:51.460 People enjoy it.

00:06:52.720 They enjoy the feeling of kindness towards other people, and it makes them nicer.

00:06:56.080 And recent studies, like very recent studies, by the psychologist David Desteno in Northwestern,

00:07:02.760 back this up by finding that meditation training actually increases people's kindness.

00:07:07.900 And the explanation that they give – and it's an open question why it does so – the

00:07:11.460 explanation they give is, it ignites compassion but shuts down empathy circuitry.

00:07:15.880 That is, you deal with suffering, and you could deal with it better because you don't

00:07:20.400 feel it.

00:07:21.260 So, this is one way I'd make the distinction between empathy and compassion.

00:07:24.340 Yeah, I think we probably raised this last time, but it's difficult to exaggerate how

00:07:30.220 fully our moral intuitions can misfire when guided by empathy as opposed to some kind of

00:07:35.620 rational understanding of what will positively affect the world.

00:07:40.220 The research done by Paul Slovic on moral illusions is fascinating here.

00:07:46.620 When you show someone a picture of a single little girl who's in need, they are maximally 0.99

00:07:52.120 motivated to help.

00:07:54.340 But if you show them a picture of the same little girl and her brother, their altruistic

00:08:00.220 motive to help is reduced reliably.

00:08:02.820 And if you show them 10 kids, it's reduced further.

00:08:05.620 And then if you give them statistics about hundreds of thousands of kids in need of this

00:08:09.720 same aid, it drops off a cliff.

00:08:12.340 And that is clearly a bug, not a feature.

00:08:14.960 And that, I think, relates to this issue of empathy as opposed to what is a higher cognitive act of

00:08:23.400 just assessing where the needs are greatest in the world.

00:08:27.100 One could argue that we are not evolutionarily well designed to do that.

00:08:31.620 We aren't.

00:08:32.800 I mean, I remember you cited the Slovic findings.

00:08:35.000 I think it was in the moral landscape where you say something to the fact that there's

00:08:40.060 never been a psychological finding that so blatantly shows a moral error.

00:08:44.340 Whatever your moral philosophy is, you shouldn't think that one life is worth more than eight,

00:08:50.360 let alone worth more than 100.

00:08:51.860 Especially when the eight contain the one life you're concerned about.

00:08:55.160 Exactly.

00:08:55.720 Yeah.

00:08:56.040 Exactly.

00:08:56.660 It's a moral disaster.

00:08:58.160 And I mean, the cool thing is that upon reflection, we could realize this. 0.99

00:09:01.980 So I'm not one of these psychologists who go on about how stupid we are, because I think 0.99

00:09:05.660 every demonstration of human stupidity or irrationality has contained with it a demonstration 1.00

00:09:11.400 of our intelligence, because we know it's irrational. 1.00

00:09:14.180 We could point it out and say, God, that's silly.

00:09:16.500 I mean, and we have a lot of, my book cites a lot of research show demonstrating the sort of

00:09:21.520 phenomena you're talking about, but it's an old observation.

00:09:24.320 I mean, Adam Smith, like 300 years ago, over, yeah, but, but 300 years ago said, gave the

00:09:30.020 example of an educated man of Europe hearing that the country of China was destroyed.

00:09:36.400 At a time when they would have never known somebody from China.

00:09:39.140 And Smith says, basically, your average European man would say, well, that's a shame.

00:09:44.140 And he'd go on his day.

00:09:45.720 But if he was to learn that tomorrow, he would lose his little finger.

00:09:49.220 He'd freak out.

00:09:50.020 He wouldn't sleep at all at night.

00:09:52.280 How am I losing my fingers?

00:09:53.300 Will it be painful?

00:09:54.200 Will it be, how will it affect my life?

00:09:56.580 And he uses this example to show that our feelings are skewed in bizarre ways.

00:10:01.660 But then he goes on to point out that we can step back and recognize that the death of thousands

00:10:07.000 is far greater tragedy than the loss of our finger.

00:10:10.740 And it's this dualism, this duality that fascinates me between what our gut tells us and what our

00:10:16.840 minds tell us.

00:10:17.660 I believe he also goes on to say that any man who would weigh the loss of his finger over

00:10:22.820 the lives of thousands or millions in some distant country, we would consider a moral

00:10:27.840 monster.

00:10:28.300 Yes, he says that human nature shudders at the thought.

00:10:32.220 Right.

00:10:32.620 It's one of the great passages in all of literature, really.

00:10:35.840 I think I quote the whole thing in The Moral Landscape.

00:10:38.660 So just a few points to pick up on what you just said about the neuroimaging research done

00:10:46.040 on empathy versus compassion.

00:10:47.940 It's something that people don't tend to know about the meditative side of this.

00:10:52.920 But compassion as a response to suffering from a meditative first person, and certainly

00:11:00.480 from the view of Buddhist psychology, is a highly positive emotion.

00:11:06.120 It's not a negative emotion.

00:11:07.380 You're not diminished by the feeling of compassion.

00:11:10.180 The feeling of compassion is really exquisitely pleasurable.

00:11:14.260 It is what love feels like in the presence of suffering.

00:11:18.100 The Buddhists have various modes of what is called loving-kindness, and loving-kindness

00:11:24.900 is the generic feeling of wishing others happiness.

00:11:29.680 And you can actually form this wish with an intensity that is really psychologically overwhelming,

00:11:37.160 which is to say it just drowns out every other attitude you would have toward friends

00:11:43.200 or strangers or even enemies.

00:11:44.960 You can just get this humming even directed at a person who has done you harm or who is

00:11:52.160 just kind of objectively evil.

00:11:54.040 You wish this person was no longer suffering in all the ways that they are and will to be

00:12:00.500 the kind of evil person they are, and you wish you could improve them.

00:12:04.640 And so Buddhist meditators acquire these states of mind, and it's the antithesis of merely being 0.98

00:12:12.900 made to suffer by witnessing the suffering of others.

00:12:17.120 It's the antithesis of being made depressed when you are in the presence of a depressed

00:12:22.460 person, say.

00:12:23.420 And so it really is the fact that empathy and compassion are used, for the most part, as

00:12:29.260 synonyms in our culture is deeply confusing about what normative human psychology promises

00:12:38.780 and just what is on the menu as far as conscious attitudes one can take toward the suffering

00:12:45.420 of other people.

00:12:45.980 I think that's right.

00:12:47.560 I think that I'm now in a sort of getting in a debate in the journal Trends in Cognitive

00:12:53.180 Sciences with an excellent neuroscientist who disagrees with me.

00:12:57.700 And there's all sorts of interesting points to go back and forth.

00:13:00.240 But at one point, he complains about the terminology, and he says, compassion isn't opposed to empathy.

00:13:05.120 It's a type of empathy.

00:13:06.700 To which my response is, who cares?

00:13:08.460 I don't care how one calls it.

00:13:11.540 I'm totally comfortable to call them different types of empathy, in which case I'm against

00:13:15.360 one type of empathy before another.

00:13:17.620 But the distinction itself is absolutely critical.

00:13:20.980 And it's so often missed, not only in the scientific field, but also in everyday life.

00:13:26.480 I published an article on empathy in the Boston Review, and I got a wonderful letter, which

00:13:30.680 I quote in my book, with permission of the writer, by this woman who worked as a first

00:13:36.820 responder after 9-11.

00:13:39.140 And after doing this for about a week, she couldn't take it anymore.

00:13:42.040 She was too oppressed by what she felt.

00:13:44.700 While her husband happily and cheerfully continued his work, and it didn't seem to harm him at

00:13:51.820 all.

00:13:52.100 And she was like, what is going on?

00:13:53.660 Is something wrong with him?

00:13:54.700 Is something wrong with me?

00:13:55.940 And I think we make sense of this by saying that there's at least two processes that lead

00:14:02.340 to kindness and good behavior that can.

00:14:04.880 And one of them, empathy, has some serious problems.

00:14:07.520 And if we could nurture compassion, we not only can make the world a better place, but

00:14:10.900 we could enjoy ourselves while doing it.

00:14:13.960 To be clear, you also differentiate two versions of empathy, because there is the cognitive

00:14:18.940 empathy of simply understanding another person's experience.

00:14:23.400 And then there's the emotional contagion version, which we're talking about, which is you are permeable

00:14:29.080 to their suffering in a way that makes you suffer also.

00:14:32.360 So that's right.

00:14:33.220 The cognitive empathy is kind of a different bag, and it's very interesting.

00:14:36.680 And we might turn to this later if we talk about Trump, but it's an understanding of what

00:14:41.680 goes on in the minds of other people.

00:14:43.720 And sometimes we call this mind reading or theory of mind or social intelligence.

00:14:48.240 And to me, it's neither good nor bad.

00:14:49.880 It's a tool.

00:14:50.440 If you, Sam, want to make the world a better place and help people, help your family, help

00:14:55.700 others, you can't do it unless you understand what people want, what affects people, what

00:15:00.300 people's interests are, what they believe.

00:15:03.460 Any good person, any good policymaker needs to have high cognitive empathy.

00:15:07.800 On the other hand, suppose you wanted to bully and humiliate people, to seduce them against

00:15:12.540 their will, to con them, to torture them.

00:15:14.320 Here, too, high cognitive empathy will help.

00:15:18.380 If you want to make me miserable, it really helped to know how I work and how my mind works.

00:15:23.960 So cognitive empathy is a form of intelligence, like any sort of intelligence can be used in

00:15:28.340 different ways.

00:15:29.440 It's morally neutral.

00:15:30.400 So to say that someone is highly empathic in that way is to simply say that they can take

00:15:35.600 another person's point of view, but that can be used for good or evil.

00:15:40.280 That's right.

00:15:40.740 The worst people in the world have high cognitive empathy.

00:15:44.240 It's how they're able to do so much damage.

00:15:46.060 Right.

00:15:46.660 I wanted to step back to something you said about meditative practice and Buddhism, because

00:15:50.660 there were two things you said, and one is easy really to get behind, which is the pleasure

00:15:54.940 that comes through this sort of practice in doing good, in loving people, in caring about

00:16:00.080 people.

00:16:00.920 But one thing I struggle with, and I don't know whether we have different views on this, is

00:16:06.160 over the blurring of distinctions that comes through Buddhism in this meditative practice.

00:16:12.300 So there's a joke I like.

00:16:14.520 It's my only Buddhism joke. 0.98

00:16:16.360 Have you heard about the Buddhist vacuum cleaner?

00:16:18.500 It comes with no attachments.

00:16:20.540 And so one of the criticisms of Buddhist practice, and to some extent a criticism of some of my

00:16:26.800 positions, is that there's some partiality we do want to have.

00:16:31.620 If I'm not only do I love my children more than, well, more than I love you, but I think

00:16:37.060 I'm right to love my children more than I love you.

00:16:40.260 Okay.

00:16:40.340 We're going to end the podcast right now.

00:16:43.200 One of the requirements of my podcast guests is that they love me as much as their own children.

00:16:47.760 I love you second.

00:16:49.240 It's my two children, then you, then my wife.

00:16:52.380 Okay.

00:16:52.780 That's the appropriate ranking.

00:16:54.240 Is that good?

00:16:55.120 Especially for a third time guest.

00:16:56.960 Yes.

00:16:57.280 I think I'm agnostic as to whether one or the other answers is normative here, or whether

00:17:03.200 there are equivalent norms, which are just mutually incompatible, but you could create

00:17:08.520 worlds that are equally good by each recipe.

00:17:12.420 But I share your skepticism, or at least it's not intuitively obvious to me, that if you could

00:17:20.260 love everyone equally, that would be better than having some gradations of moral concern.

00:17:29.380 When we extend the circle in the way that Peter Singer talks about of our moral concern,

00:17:33.780 the world does get better and better.

00:17:35.600 We want to overcome our selfishness, our egocentricity, our clannishness, our tribalism, our nationalism, 0.88

00:17:44.720 all of those things, all of those boundaries we erect where we care more about what's inside

00:17:51.320 than outside the boundary.

00:17:53.120 Those all seem, at least they tend to be pathological, and they tend to be sources of conflict, and they

00:17:59.340 tend to explain the inequities in our world that are just, on their face, unfair, and in

00:18:08.860 many cases, just unconscionable.

00:18:11.100 But whether you want to just level all of those distinctions and love all homo sapiens

00:18:17.860 equivalently, that I don't know.

00:18:21.000 And I'm not actually skeptical that it is a state of mind that's achievable.

00:18:27.400 I've met enough long-term meditators, and I've had enough experience in meditation and with

00:18:34.280 psychedelics and just changing the dials on conscious states to believe that it's possible

00:18:42.540 to actually obviate all of those distinctions and to just feel that love is nothing more

00:18:50.980 than a state of virtually limitless aspiration for the happiness of other conscious creatures

00:18:59.160 and that it need not be any more preferential or directed than that.

00:19:06.520 When you're talking about a monk who has come out of a cave doing nothing but compassion meditation

00:19:11.600 for a decade, you're talking about somebody who, in most cases, has no kids, doesn't have

00:19:17.880 to function in the world the way we have to function.

00:19:21.160 Certainly, civilization doesn't depend on people like that forming institutions and

00:19:27.280 running our world.

00:19:29.420 And so, you know, I don't know what to make of the fact that let's just grant that it's

00:19:33.320 possible to change your attitude in such a way that you really just feel equivalent love

00:19:39.140 for everybody, and there's no obvious cost to you for doing that.

00:19:44.080 I don't know what the cost would be to the species or to society if everyone was like

00:19:49.500 that.

00:19:49.880 And intuitively, I feel like it makes sense for me to be more concerned and therefore much

00:19:55.560 more responsible for and to my kids than for yours.

00:20:00.820 But at a greater level of abstraction, when I talk about how I want society to be run, I

00:20:07.000 don't want to try to legislate my selfishness.

00:20:09.540 I just have to understand that at the level of laws and institutions, fairness is a value

00:20:16.440 that more often than not conserves everyone's interests better than successfully gaming a

00:20:23.360 corrupt system.

00:20:24.600 Yeah.

00:20:24.880 So I'm nodding.

00:20:26.040 I mean, I want to zoom in on the last thing you said because it was astonishing to me.

00:20:29.680 But for most of what you're saying, I'm nodding in agreement.

00:20:32.980 Certainly, the world would be much better if our moral circle was expanded.

00:20:38.600 And certainly, the world would be much better if we cared a little bit more for people outside

00:20:44.480 of our group and correspondingly, relatively less for people inside of our group.

00:20:49.080 It's not that we don't care for our own enough.

00:20:52.000 The problem is we don't care for others enough.

00:20:54.380 And I love your distinction as well, which is a way I kind of think about it now is, yeah,

00:20:59.640 I love my children more than I love your children.

00:21:02.920 But I understand stepping back that a just society should treat them the same.

00:21:09.940 Right.

00:21:10.060 So if I have a summer job opening, I understand my university regulations say I can't hire

00:21:17.300 my sons.

00:21:18.680 And, you know, I actually think that's a good rule.

00:21:20.600 I love my sons.

00:21:21.520 I'd like to hire them, a job for them and everything.

00:21:24.000 But I could step back and say, yeah, we shouldn't be allowed to let our own personal preferences,

00:21:31.060 our own emotional family ties distort systems that should be just and fair.

00:21:36.640 The part of what you said, which I just got to zoom in on, is do you really think it's

00:21:41.500 possible, put aside somebody who has no prior attachments at all, some monk living in a

00:21:46.840 cave, have you met or do you think you will ever meet people who have had children and

00:21:52.540 raised them who would treat the death of their child no differently than the death of a strange 0.96

00:21:59.000 child?

00:21:59.700 Yeah, I do, actually.

00:22:01.740 Wow.

00:22:02.160 And I'm not sure what I think of it.

00:22:05.540 I can tell you these are extraordinarily happy people.

00:22:09.180 So what you get from them is not a perceived deficit of compassion or love or engagement

00:22:16.540 with the welfare of other people, but you get a kind of obliteration of preference.

00:22:24.920 The problem in their case is it's a surfeit of compassion and love and engagement so that

00:22:31.900 they don't honor the kinds of normal family ties or preferences that we consider normative

00:22:40.220 and that we would be personally scandalized to not honor ourselves.

00:22:44.700 The norms of preference, which seem good to us and we would feel that we have a duty to

00:22:51.460 enforce in our own lives and we would be wracked by guilt if we noticed a lapse in honoring

00:22:57.260 those duties, these are people who have just blown past all of that because they have used

00:23:02.540 their attention in such an unconventional and in some ways impersonal way, but it's an

00:23:08.360 impersonal way that becomes highly personal or at least highly intimate in their relations

00:23:13.500 with other people.

00:23:14.320 So for instance, I studied with one teacher in India, a man named Poonjaji.

00:23:20.640 He actually, he wasn't Buddhist.

00:23:21.900 He was Hindu, but he was not teaching anything, especially Hindu. 0.92

00:23:26.840 I mean, he was talking very much in the tradition of, if people are aware of these terms and I'll

00:23:32.060 get them from my book, Waking Up, the tradition of Advaita Vedanta, the non-dual teachings of

00:23:36.980 Vedanta, which are nominally Hindu.

00:23:39.000 They're really just Indian and there's nothing about gods or goddesses or any of the garish 0.99

00:23:43.220 religiosity you see in Hinduism.

00:23:46.100 He was a really, I mean, there was a lot that he taught that I disagreed with and, or at least

00:23:52.180 there were some crucial bits that he taught that I disagreed with.

00:23:55.000 And again, you can find that in Waking Up if you're interested, but he was a really shockingly

00:24:02.220 charismatic and wise person to be in the presence of.

00:24:08.160 He was really somebody who could just bowl you over with his compassion and his, the force

00:24:16.940 of his attention.

00:24:18.360 If I were not as scrupulous as I am about attributing, you know, causality here, 90% of the people

00:24:26.840 who spent any significant time around this guy thought he had, you know, magic powers.

00:24:33.000 This is a highly unusual experience of being in a person's presence.

00:24:38.080 Part of what made him so powerful was that actually, ironically, he had extraordinarily high

00:24:45.000 empathy of the unproductive kind, but it was kind of anchored to nothing in his mind.

00:24:52.060 So, for instance, if someone would have a powerful experience in his presence and, you know,

00:24:57.680 start to cry, you know, tears would just pour out of his eyes.

00:25:01.940 You know, he would just immediately start crying with the person.

00:25:05.160 And when somebody would laugh, he would laugh, you know, twice as hard.

00:25:08.720 It was like he was a amplifier of the states of consciousness of the people around him in

00:25:14.920 a way that was really thrilling.

00:25:17.020 And, again, there was, you know, a feedback mechanism here where, you know, people would

00:25:21.080 just have a bigger experience because of the ways in which he was mirroring their experience.

00:25:25.520 And there was no sense at all that this was an act.

00:25:28.220 I mean, he would have to have been the greatest actor on earth for this to be brought off.

00:25:32.960 But, yeah, he's, I think, I forget the details of the story, but the story about, you know,

00:25:37.700 how he behaved when his own son died would shock you with its apparent aloofness, right?

00:25:44.900 I mean, this is a person for whom a central part of his teaching was that death is not

00:25:49.420 a problem and he's not hanging on to his own life or the lives of those he loves with any

00:25:55.140 attachment at all.

00:25:56.220 And he was advertising the benefits of this attitude all the time because he was the happiest

00:26:02.180 man you ever met.

00:26:03.260 But I think when push came to shove and he had to react to the death of his son, he wouldn't

00:26:08.980 react the way you or I would or the way you or I would want to react given how we view

00:26:15.000 the world.

00:26:15.660 That's an extraordinary story.

00:26:16.960 I mean, I, you have a lot of stories like that and waking up of people like that.

00:26:21.500 And I haven't encountered many such people.

00:26:23.960 I met Matthew Ricard once and it was a profoundly moving experience for me.

00:26:28.420 And I'm not, I'm not easily impressed.

00:26:30.700 I'm sort of, I tend to be cynical about people.

00:26:32.640 I tend to be really cynical about people who claim to have certain abilities and the

00:26:36.780 like, but I simply had a short meeting from a man out for tea and we just talked and there's

00:26:42.100 something about people who have attained a certain mental capacity or set of capacities

00:26:47.660 that you can tell by being with them that they have it, their bodies afford it.

00:26:52.460 They just, they just give it off from a mile away.

00:26:56.020 It's like, um, it's, it's, it's analogous to charisma, which some people have apparently.

00:27:02.660 Um, Bill Clinton is supposed to be able to walk into like a, a, a large room and people

00:27:07.600 just gather around him.

00:27:09.220 Their eyes will be drawn towards him.

00:27:10.700 And whatever it is that someone like Matthew Ricard has is, is extraordinary in a different

00:27:15.600 way, which is he has some literal sense exudes peace and compassion.

00:27:20.540 Having said that, um, some of it freaks me out and some of it morally troubles me.

00:27:25.300 I mean, we talked about the bonds of family, but I can't imagine any such people having

00:27:29.680 bonds of friendship.

00:27:31.260 Um, you, I would imagine you get a lot of email, Sam.

00:27:34.740 I imagine you get a lot of email asking you for favors.

00:27:37.420 So when I email you and say, Hey, you know, I have a book coming out.

00:27:40.480 You're in a mood for a podcast.

00:27:42.060 You, you, um, because we're friends, you respond to me different than if I were a total

00:27:46.440 stranger.

00:27:47.500 Suppose you didn't suppose you treated everything on its merits with no bonds, no, no connectedness.

00:27:54.140 It'd be hard to see you as a good person.

00:27:56.700 But you don't see Matthew that way.

00:27:59.780 No.

00:28:00.180 If you knew more about the details of his life, you might find that it's not aligned

00:28:05.620 with the way you parcel your concern.

00:28:08.840 For instance, the example you just gave, he might be less preferential toward friends or

00:28:14.340 not.

00:28:14.520 I actually know Matthew.

00:28:15.500 I don't often see him, but I've spent a fair amount of time with him.

00:28:19.620 I mean, he's what I would call a mensch. 1.00

00:28:21.740 He's just like the most decent guy you're going to meet all year.

00:28:25.980 He's just a, he's just a wonderful person, but he's a, I studied with his teacher, Kensei

00:28:31.780 Rinpoche, who was a very famous lama and who many, you know, Tibetan lamas thought was,

00:28:37.540 you know, one of the greatest meditation masters of his generation.

00:28:40.440 He died, unfortunately, about 20 years ago, but maybe it's more than that.

00:28:44.300 And I now notice as I get older, whenever I estimate how much time has passed, I'm off

00:28:49.500 by 50% at least.

00:28:51.660 Someone should study that problem.

00:28:53.260 Self-deception, I think, has something to do with it.

00:28:55.180 So anyway, Kensei Rinpoche was just this 800-pound gorilla of meditation.

00:29:00.840 He'd spent more than 20 years practicing in solitude.

00:29:04.660 And Matthew was his closest attendant for years and years.

00:29:10.920 I think just to give you kind of to rank order what's possible here, Matthew certainly wouldn't

00:29:15.600 put himself anywhere near any of his teachers on the hierarchy of what's possible in terms

00:29:21.660 of, you know, transforming your moment-to-moment conscious experience and therefore the likelihood

00:29:26.660 of how you show up for others.

00:29:29.300 Matthew's great because, as you know, he's got this, he was a scientist before he became

00:29:33.640 a monk, he was a molecular biologist.

00:29:36.240 And the work he's done in collaborating with neuroscientists who do neuroimaging work on

00:29:42.720 meditation has been great.

00:29:45.300 And he's, you know, he's a real meditator, so he can honestly talk about what he's doing

00:29:50.600 inside the scanner, and that's fantastic.

00:29:53.120 But again, even in his case, he's made a very strange life decision, certainly from your

00:30:00.540 point of view.

00:30:01.000 He's decided to be a monk and to not have kids, to not have a career in science, to

00:30:06.640 not, it's in some ways an accident that he, that you even know about him because he could

00:30:12.320 just be, and for the longest time he was just sitting in a tiny little monk cell in

00:30:18.220 Kathmandu serving his guru.

00:30:21.360 That's right.

00:30:21.620 And when I met him, he was spending six months of each year in total solitude, which again,

00:30:26.380 boggles my mind, because if I spend a half hour by myself, I start to want to check my

00:30:30.620 email and get anxious.

00:30:31.640 And I accept your point, which is, I need to sort of work to become more open-minded about

00:30:41.380 what the world would be like if certain things which I hold dear were taken away.

00:30:45.420 There's a story I like of why economics got called dismal science.

00:30:50.700 And it's because the term was given by Carlyle, and Carlyle was enraged at the economists who

00:30:57.360 were dismissing an institution that Carlyle took very seriously.

00:31:02.100 And the economist said, this is an immoral institution, and Carlyle says, you have no

00:31:05.840 sense of feeling, you have no sense of tradition.

00:31:08.560 And he was talking about slavery.

00:31:09.640 And so, you know, he was blasting the economist for being so cold-blooded, they couldn't appreciate

00:31:14.740 the value and importance of slavery.

00:31:16.580 And sometimes when I feel my own emotional pulls towards certain things, and I feel like,

00:31:21.900 I feel confident that whatever pulls I have along, say, racial lines are immoral, but 1.00

00:31:27.440 I'm less certain about family lines or friendship lines, I think I need to be reminded, we all

00:31:32.340 need to be reminded, well, we need to step back and look, what will future generations say?

00:31:36.860 What will we say when we're at our best selves?

00:31:39.640 It's going to take more than that for me to give up the idea that I should love my children

00:31:43.520 more than I love your children, but it is worth thinking about.

00:31:47.080 And it's interesting to consider moral emergencies and how people respond in them and how we would

00:31:53.920 judge their response.

00:31:55.040 So just imagine if, you know, you had a burning building and our children were in there and

00:32:02.000 I could run in to save them, say, I'm on site and I can run in and save whoever I can

00:32:09.300 save, but because I know my child's in there, my priority is to get my child and who could

00:32:15.660 blame me for that, right?

00:32:17.040 That's right.

00:32:17.380 So I run in there and I see your child who I can quickly save, but I need to look for

00:32:23.000 my child.

00:32:23.620 So I just run past your child and go looking for mine.

00:32:27.940 And at the, you know, the end of the day, I save no one, say, or I only save mine.

00:32:33.760 It really, really was a zero sum contest between yours and mine.

00:32:37.240 You know, if you could watch that play out, if you had a video of what I did in that house,

00:32:41.280 right?

00:32:41.580 And you saw me run past your child and just go looking for mine, I think it's just hard

00:32:47.680 to know where to find the norm there.

00:32:52.580 A certain degree of searching and a certain degree of disinterest with respect to the fate

00:32:58.880 of your child begins to look totally pathological.

00:33:02.760 It looks just morbidly selfish.

00:33:04.680 But some bias seems only natural and we might view me strangely if I showed no bias at all.

00:33:12.740 Again, I don't know what the right answer is there.

00:33:15.280 We're living as though almost a total detachment from other people's fates, apart from the fates

00:33:21.480 of our own family, is normal and healthy.

00:33:25.600 And when push comes to shove, I think that is clearly revealed to not be healthy.

00:33:31.680 Right.

00:33:31.900 It's plain that the extremes are untenable.

00:33:33.960 Imagine you weren't looking for your child, but your child's favorite teddy bear.

00:33:38.300 Well, then you're kind of a monster, you know, searching around for that while my child 0.99

00:33:42.600 burns to death. 0.99

00:33:43.800 I mean, to make matters worse, I mean, Peter Singer is famously, I think he very convincingly

00:33:48.140 pointed out that the example you're giving is a sort of weird science fiction example and

00:33:52.740 you might reassure, we might reassure, says, well, that'll never happen.

00:33:56.000 But Singer points out we're stuck in this dilemma every day of our lives.

00:34:00.320 Yeah.

00:34:00.520 As we devote resources, you know, I, like a lot of parents, spend a lot of money on my

00:34:06.020 kids, including things that they don't, you know, things that make your lives better but

00:34:10.120 aren't necessary.

00:34:11.160 And things that are just fun, expensive toys and vacations and so on, while other children

00:34:15.960 die.

00:34:16.300 And, and Singer points out that, um, I really am in that burning building.

00:34:21.480 Yeah.

00:34:21.920 I am in that burning building buying my son an Xbox while kids from Africa die in the 1.00

00:34:27.780 corner.

00:34:28.760 And, and it's, it's difficult to confront this.

00:34:31.300 And I think people get very upset when Peter Singer rings it up, rings it up.

00:34:34.980 But it is a moral dilemma that we are continually living with and continually struggling with.

00:34:41.140 And I don't know what the right answer is, but I do have a sense that the way we're doing it

00:34:45.540 every day is the wrong way.

00:34:47.980 We are not devoting enough attention to those in need.

00:34:50.440 We're devoting too much attention to those we love.

00:34:53.100 Yeah.

00:34:53.440 Yeah.

00:34:53.580 Well, I had Singer on the podcast.

00:34:55.440 I also had Will McCaskill, who's very much argues in the same line.

00:34:59.120 And, you know, I don't have a good answer.

00:35:02.460 I think one thing I did as a result of my conversation with Will was I realized that I

00:35:08.240 just, I needed to kind of automate this insight.

00:35:11.420 So Will is very involved in the, the effect of altruism community.

00:35:15.860 And he arguably, I think, started the movement and their websites like givewell.org that rate

00:35:21.780 charities.

00:35:23.100 And they've quantified that to, to save a, an individual human life costs now $3,500.

00:35:29.240 I mean, that's, that's the amount of money you have to allocate where you can say as,

00:35:33.860 as a matter of likely math, you have saved one human life.

00:35:38.200 And the calculation there is, is with reference to the work of the Against Malaria Foundation.

00:35:43.260 They, they put up these insecticide treated bed nets and malaria death has come down by

00:35:48.860 50%.

00:35:49.820 It's still close to a million people dying every year, but it was very recently 2 million people

00:35:54.000 dying a year from mosquito-borne, not, not, not all mosquito-borne illness, just malaria,

00:35:58.680 actually.

00:35:59.240 So in response to my conversation with Will, I just decided, well, I'm still going to buy

00:36:03.980 the Xbox.

00:36:04.800 I know, I know that I can't conform my life and my, you know, the fun I have with my kids

00:36:10.500 so fully to this logic of this triage, right?

00:36:15.200 So that I, you know, strip all the fun out of life and just give everything to the Against

00:36:19.940 Malaria Foundation.

00:36:20.800 But I decided that the first $3,500 that comes into the podcast every month will just by definition

00:36:27.680 go to the Against Malaria Foundation.

00:36:29.500 I don't have to think about it.

00:36:30.600 It just happens every month.

00:36:32.020 I would have to decide to stop it from happening.

00:36:35.140 I think doing more things like that.

00:36:37.200 I mean, so what Will does is there's actually a giving pledge where people decide to give

00:36:41.880 10% of their, I think it's at least 10% of their income to charity and to these most effective

00:36:48.200 charities each year, any kind of change you want to see in the world that you want to

00:36:53.500 be effective, automating it and taking it out of the cognitive overhead of having to

00:36:59.940 be re-inspired to do it each day or each year or each period, that's an important thing

00:37:05.500 to do.

00:37:06.240 That's why I think the greatest changes in human well-being and in human morality will

00:37:11.860 come not from each of us individually refining our ethical code to the point where we are

00:37:18.780 bypassing every moral illusion, right?

00:37:20.980 So that every time Paul Slovic shows us a picture of a little girl, we have the exact

00:37:25.540 right level of concern.

00:37:26.880 And when we see eight kids, we have, you know, we have eightfold more or whatever it would

00:37:31.080 be.

00:37:31.360 But to change our laws and institutions and tax codes and everything else so that more good

00:37:37.900 is getting done without us having to be saints in the meantime.

00:37:40.820 I think that that's right.

00:37:42.040 I think that this comes up a lot in discussions of empathy.

00:37:45.020 So I, you know, I talk about the failings of empathy in our personal lives, particularly

00:37:49.580 say giving to charity or deciding how to treat other people.

00:37:52.680 And a perfectly good response I sometimes get is, well, okay, I'm a high empathy person.

00:37:57.140 What am I going to do about it?

00:37:58.760 And, you know, one answer concerns activities like meditative practice.

00:38:02.900 But, you know, you could be skeptical over how well that works for many people.

00:38:06.760 Um, I mean, I think your answer is best, which is in a good society and actually as good

00:38:12.020 individuals, we're smart enough to develop procedures, uh, mechanisms that take things

00:38:18.780 out of our hands.

00:38:19.860 And this applies at every level.

00:38:22.060 Uh, the political theorist, Jan Elster points out, that's what a constitution is.

00:38:26.320 A constitution is a bunch of people saying, look, we are irrational people. 0.95

00:38:30.440 And sometime in the future, we're going to be tempted to make dumb, irrational choices. 0.93

00:38:34.960 So let's set up something to stop us.

00:38:37.780 And let's set us, let's set up something that, um, to override our base instincts.

00:38:42.940 We can change this, this stopping mechanism, but let's make it difficult to change.

00:38:47.920 So it takes years and years.

00:38:49.380 So no matter how much Americans might choose, they want to reelect a popular president for

00:38:54.680 a third term.

00:38:55.480 They can't.

00:38:56.440 If all the white Americans decide they want to re-instantiate the institution of slavery, 0.75

00:39:00.860 they can't.

00:39:02.320 And laws work that way.

00:39:03.900 Constitutions work that way.

00:39:05.020 I think, I think good diets work that way.

00:39:07.200 Yeah.

00:39:07.700 And, and charitable giving could work that way, um, in that you have, uh, automatic withdrawal

00:39:12.940 or whatever.

00:39:13.620 So you, in an enlightened moment, you say, this is the kind of person I want to be.

00:39:17.500 And you don't wait for your gut feelings all the time.

00:39:20.880 I think, um, overriding other, uh, disruptive sentiments works the same way.

00:39:26.820 Like, um, suppose I have to choose somebody to be a graduate student or, or something like

00:39:31.720 that.

00:39:32.440 And I know full well that there are all sorts of biases having to do with physical attractiveness,

00:39:37.900 with race, with gender.

00:39:39.820 And suppose I believe, upon contemplation, that it shouldn't matter.

00:39:44.200 It shouldn't matter how good looking the person is.

00:39:45.920 It shouldn't matter whether they were from the same country as me.

00:39:48.860 Well, one thing I could try to do is say, okay, I'm going to try to really be very hard.

00:39:53.320 I'm going to try very hard not to be biased.

00:39:55.660 We're horrible at that.

00:39:56.940 Yeah.

00:39:57.380 We overcorrect.

00:39:58.320 We undercorrect.

00:39:59.120 We justify.

00:39:59.740 So what we, what we do when we're at our best is develop some systems.

00:40:04.760 Like, for instance, you, um, you don't look at the pictures.

00:40:07.340 You do some sort of blind, blind reviewing.

00:40:10.040 So your biases can't come into play.

00:40:12.080 Yeah.

00:40:12.600 Now it's harder to see how this is done when it comes to broader policy decisions, but

00:40:18.420 people are working on it.

00:40:19.520 Paul Slovic actually, who we've referenced a few times, talks about this a lot.

00:40:23.180 So right now, for instance, government's decisions over where to send aid or where to go to war

00:40:30.080 are made basically on gut feelings.

00:40:32.480 And they're basically based on sad stories and photographs of children washed ashore and

00:40:37.660 so on.

00:40:38.280 And it just makes the world worse.

00:40:41.080 And people like Slovic wonder, can we set up some fairly neutral triggering procedures

00:40:46.360 that say in America, when a situation gets this bad, according to some numbers and some

00:40:51.780 objective judgments, it's a national emergency.

00:40:53.860 We send in money, uh, overseas.

00:40:55.940 If this many people die under such and so circumstances, we initiate some sort of investigative

00:41:00.800 procedure.

00:41:01.520 It sounds cold and bureaucratic, but I think cold and bureaucratic is much better than hot

00:41:07.440 blooded and arbitrary.

00:41:09.180 There was something you said when referencing the soccer study in group empathy and out group

00:41:15.140 schadenfreude, I guess.

00:41:16.800 Yes.

00:41:17.280 And this was a, this reminded me of a question we got on Twitter.

00:41:20.200 Someone was asking about the relationship between empathy and identity politics.

00:41:24.920 I guess, I guess based on the research you just cited, there's a pretty straightforward

00:41:28.680 connection there.

00:41:30.060 You have any thoughts on that?

00:41:31.280 There's a profound connection.

00:41:33.180 We're, we're very empathic creatures, but it always works out that the empathy tends to,

00:41:39.720 to focus on those from within our group and not to the out group.

00:41:43.280 I got into a good discussion once with Simon Baron Cohen, the psychologist who's very pro

00:41:47.160 empathy.

00:41:47.500 And he said that if only, um, we're talking about his time of the, of the, uh, war in

00:41:53.320 Gaza.

00:41:53.960 And, um, he's talking to only the Palestinians and Israelis had more empathy.

00:41:57.840 This wouldn't happen.

00:41:59.160 The Israelis would realize that, uh, the suffering of the Palestinians and vice versa, and there'd

00:42:04.100 be peace.

00:42:04.940 And, and my feeling here is that that's exactly, it's exactly the opposite.

00:42:09.160 That, that conflict in particular suffered from an abundance of empathy.

00:42:14.100 The Israelis at the time felt huge empathy for the suffering of teenagers who were kidnapped

00:42:19.140 of their families.

00:42:20.960 The Palestinians felt tremendous empathy for their countrymen who were imprisoned and tortured.

00:42:26.020 Um, there were abundant empathy and there's always abundant empathy at the core of any

00:42:31.080 conflict.

00:42:31.580 And the reason why it drives conflict is I feel tremendous empathy for the American who is

00:42:37.300 tortured and captured.

00:42:38.240 And as a rule, it's very hard for me to feel empathy for the Syrian or for the Iraqi and

00:42:44.180 so on.

00:42:45.200 And, and, you know, we could, we could now pull it down a little bit in the aftermath of

00:42:49.720 the 2016 election.

00:42:51.760 I think, um, I think Clinton voters are exquisitely good at empathy towards other Clinton voters

00:42:56.960 and Trump voters for Trump voters.

00:42:58.680 Having empathy for your political enemies is, is difficult.

00:43:02.240 And I think actually, and for the most part, um, so hard that it's not worth attaining, we

00:43:07.680 should try for other things.

00:43:08.980 I think we certainly want the other form of empathy.

00:43:11.880 I mean, we want to be able to understand why people decided what they decided.

00:43:16.460 And we don't want to be just imagining motives that don't exist or, or weighting them in ways

00:43:21.640 that are totally unrealistic.

00:43:24.180 We inevitably will say something about politics.

00:43:26.760 People would expect that of us.

00:43:28.320 By, by law, there could be no discussion of over 30 minutes that doesn't, uh, mention

00:43:32.100 Trump.

00:43:32.920 I'm going to steer you toward Trump, but before we go there, as you may or may not know, I've

00:43:37.600 been fairly obsessed with artificial intelligence over the last, I don't know, 18 months or so.

00:43:43.760 And we solicited some questions from Twitter and many people asked about this.

00:43:49.200 Have you gone down this rabbit hole at all?

00:43:51.160 Do you, have you thought about AI much?

00:43:53.460 And there was one question I saw here, which was given your research on empathy, how should

00:43:59.180 we program an AI with respect to it?

00:44:03.080 So I actually hadn't taken seriously the AI worries.

00:44:06.200 Um, and honestly, I'll be honest, I dismissed them as somewhat crackpot until I listened to

00:44:11.180 you talk about it.

00:44:11.860 I think it was a TED talk.

00:44:13.360 Um, and, uh, and then, and then, so thank you.

00:44:16.880 That got me worried about yet something else.

00:44:18.540 Um, and I, I found it fairly persuasive that there's an issue here.

00:44:22.440 We should be devoting a lot more thought to, um, the question of putting empathy into machines,

00:44:29.280 which is, is, um, is I think in some way morally fraught because, um, if I'm right that

00:44:36.680 empathy leads to capricious and arbitrary decisions, then if we put empathy into computers

00:44:42.180 or robots, we end up with capricious and arbitrary computers and robots.

00:44:45.660 I think when people think about putting empathy into machines, they often think about it from

00:44:49.620 a marketing point of view, such that, um, uh, you know, a household robot or even an interface

00:44:56.200 on a Mac computer that is somewhat empathic will, um, will be more pleasurable to interact

00:45:02.620 with more humanoid, more human-like, and we'll get more pleasure, uh, dealing with it.

00:45:07.380 And that might be the case.

00:45:08.880 I, I've actually heard a Contra review, uh, from my friend, David Pizarro, who points out

00:45:13.900 that when dealing with a lot of, uh, interactions, we actually don't want empathy.

00:45:20.260 We want a sort of, uh, cold-blooded interaction that we don't have to become emotionally invested

00:45:25.240 in.

00:45:26.100 We want professionalism.

00:45:27.280 I think, I think of our super intelligent AI, I think we want professionalism more than

00:45:33.060 we want emotional contagion.

00:45:35.120 You don't want, if you're anxious and consulting your robot doctor, you don't want that anxiety

00:45:43.040 mirrored back to you, right?

00:45:44.860 You want, you want as stately a physician as you ever met in, in the living world now embodied

00:45:51.440 in this machine.

00:45:52.520 Yeah.

00:45:52.680 So I'm, I'm very happy if I have a home blood pressure cuff, which just gives me the numbers

00:45:57.080 and doesn't say, oh man, I feel terrible for you.

00:45:59.520 It doesn't make you very upsetting.

00:46:00.560 Whoa.

00:46:00.840 Yeah, yeah, dude, dude, I'm, I'm holding back here.

00:46:05.180 It's, it's, it's, it's, you know, the machine starts to a little graphical cheers trickle

00:46:09.140 down to interface.

00:46:10.440 I, I'm sure people involved in marketing these devices think that they're appealing.

00:46:14.180 I think that David is right.

00:46:16.040 And we're going to discover that for a lot of interfaces, we just want, uh, a sort of an,

00:46:21.980 an affect free, emotion free, uh, interaction.

00:46:26.400 And, um, and I think we find, um, as I find, for instance, with, with, uh, uh, interfaces

00:46:33.060 where you have to call the airport or something, when it reassures me that, that it's worried

00:46:37.380 about me and so on, I find it cloying and annoying and intrusive.

00:46:40.960 I just want, I just want the data.

00:46:43.020 Um, I, I want to, I want to save my empathy for real people.

00:46:46.160 Yeah.

00:46:46.320 But I think the question goes to what will be normative on the side of the AI?

00:46:52.660 So do we want AI, I guess, let's leave consciousness aside for the moment.

00:46:58.800 That's right.

00:46:59.120 But do we want an AI that actually has more than just factual knowledge of our preferences

00:47:08.420 insofar as it could emulate our emotional experience?

00:47:13.060 Could that give it capacities?

00:47:15.400 I see.

00:47:15.940 Yeah.

00:47:16.120 That, that we, we want it to have so as to better conserve our interests.

00:47:19.840 So here, here's what I would, here's my take on it.

00:47:22.660 I think we want AI with compassion.

00:47:25.280 I think we particularly want AI of compassion towards us.

00:47:28.680 Um, I'm not sure whether this came from you or somebody else, but somebody gave the following

00:47:32.800 scenario for how the world will end.

00:47:35.340 The world is going to end when someone programs a powerful computer that interfaces with other

00:47:40.060 things, um, to, um, to get rid of spam on email.

00:47:44.880 And then the computer will promptly destroy the world as a suitable way to do this.

00:47:48.880 Um, we want machines to be, have a guard against doing that where they say, well, human life

00:47:54.500 is valuable.

00:47:55.320 Human flourishing and animal flourishing is valuable.

00:47:58.100 So if I want, if I want AI that is involved in making significant decisions, I want it to

00:48:03.560 have compassion.

00:48:04.500 I don't, however, want it to have, uh, empathy.

00:48:08.360 I think empathy makes us, it makes us among other things, racist.

00:48:11.620 Uh, the last thing in the world we need is racist AI.

00:48:15.220 There's been some concern that we already have racist AI.

00:48:18.620 Have you heard this?

00:48:19.520 Yes, I have.

00:48:19.960 Go, go ahead.

00:48:20.760 Remind me.

00:48:21.500 If I recall correctly, there, there are algorithms that decide on the paroling of prisoners and,

00:48:28.620 or, you know, whether people get mortgages.

00:48:31.140 And there's some evidence, I could be making a, a bit of a hash of this, but there was some

00:48:36.420 evidence in, in one or both of these categories that the AI was taking racial characteristics

00:48:42.760 into its calculation.

00:48:44.140 And then that wasn't, that hadn't been programmed in, that was just an emergent property of it

00:48:49.540 finding of all the data available.

00:48:51.900 This data was, was relevant in the case of prisoners, the, the recidivism rate.

00:48:57.200 You know, if it's just a fact that black parolees recidivate more, reoffend more, I don't know

00:49:04.280 in fact that it is, but let's just say that it is.

00:49:06.460 And an AI notices that, well then of course the AI, if you're going to be predicting whether

00:49:11.740 a person is likely to violate their parole, you are going to take race into account if

00:49:16.680 it's actually descriptively true of the data, that it's a variable.

00:49:19.740 And so I think there, there was at least one story I saw where you had people scandalized

00:49:25.280 by, by racist AI.

00:49:26.700 When I was, was young and very nerdy, more nerdy than I am now, I like gobbled up all

00:49:32.440 science fiction and Isaac Asimov had a tremendous influence on me and he had all of his work

00:49:37.560 on robots and he had these three laws of robotics.

00:49:41.060 And, and, you know, if you think about it as a, you know, from a more sophisticated view,

00:49:45.180 the laws of robotics weren't particularly morally coherent.

00:49:49.100 Like one of them is you should never harm a human or through an action, allow a human to

00:49:54.320 come to harm.

00:49:54.920 But does that mean the robot's going to run around trying to save people's lives all the

00:49:59.020 time?

00:49:59.260 Because we're, we're continually not allowing people to come to harm.

00:50:02.980 But, but the spirit of the endeavor is right.

00:50:06.380 Which is, I would wire up, I think, and in fact, I think in some way as, as robots now

00:50:11.440 become more powerful, you could imagine becoming compulsory to wire up these machines with some

00:50:17.920 morality.

00:50:18.560 This comes up with driving cars, sorry, with the, with the automatic, right?

00:50:22.660 The, the, the computer driven cars where, you know, are they going to be utilitarian?

00:50:26.500 Are they going to have principles?

00:50:27.620 And there's a lot of good debates on that, but they have to be something and they have

00:50:32.740 to have some consistent moral principles that take into account human life and human flourishing.

00:50:38.200 And the last thing you want to stick in there is, is something that says, well, if someone

00:50:43.180 is adorable, care for them more.

00:50:45.140 Always count a single life as more than a hundred lives.

00:50:48.140 There's no justification for putting the sort of stupidities of empathy that we're often

00:50:52.060 stuck with to putting them into the machines that we create.

00:50:55.040 That's one thing I love about this moment of having to think about super intelligent AIs. 0.90

00:51:00.140 is it's amazingly clarifying of moral priorities.

00:51:05.740 And all these people who, until yesterday said, well, you know, who's to say what's true

00:51:10.580 in the moral sphere?

00:51:12.140 Once you force them to weigh in on how should we program the algorithm for self-driving cars,

00:51:18.980 they immediately see that, okay, you have to solve these problems one way or another.

00:51:22.520 These cars will be driving.

00:51:23.920 Do you want them to be racist cars?

00:51:25.700 Do you want them to preferentially drive over old people as opposed to children? 0.87

00:51:29.560 What makes sense?

00:51:31.220 And to say that there is no norm there to be followed is to just, you're going to be

00:51:37.720 designing one by accident then, right?

00:51:39.760 If you make a car that is totally unbiased with respect to the lives it saves, well, then

00:51:48.000 you've made this kind of this Buddhist car, right? 0.58

00:51:50.320 You've made this, you've made the Mathieu Ricard car, say.

00:51:53.380 That may be the right answer, but you have taken a position just by default.

00:51:57.900 And the moment you begin to design away from that kind of pure equality, you are forced

00:52:06.000 to make moral decisions.

00:52:07.640 And I think it's pretty clear that we have trolley problems that we have to solve.

00:52:12.160 And we have, at a minimum, we have to admit that killing one person is better than killing

00:52:16.940 five.

00:52:18.020 And we have to design our cars to have that preference.

00:52:21.520 When you put morality in the hands of the engineers, you see that you can't take refuge in any

00:52:28.740 kind of moral relativism.

00:52:30.420 You actually have to answer these questions for yourself.

00:52:33.180 I envision this future where, you know, you walk into a car dealership and you order one

00:52:37.900 of these cars and you're sitting back and you're paying for it.

00:52:40.320 And then you're asked, what kind of setting do you want?

00:52:42.260 Do you want a racist, Buddhist, radical feminist, religious fundamentalist? 1.00

00:52:46.240 I don't know if you've heard this research, but when they ask people what the cars should 0.91

00:52:50.860 do on the question of, you know, how biased should it be to save the driver over the pedestrian,

00:52:57.460 say?

00:52:57.780 So if it's a choice between avoiding a pedestrian and killing the driver or killing the pedestrian,

00:53:02.900 how should the car decide?

00:53:04.640 Most people say in the abstract, it should just be unbiased.

00:53:08.780 You should be indifferent between those two.

00:53:10.640 But when you ask people, would you buy a car that was indifferent between the driver's

00:53:15.940 life and the pedestrians, they say no.

00:53:19.020 They want a car that's going to protect their lives.

00:53:21.180 So it's hard to adhere to the thing you think is the right answer, it seems.

00:53:26.860 And there, I actually don't know how you solve that problem.

00:53:30.260 I think probably the best solution is to not advertise how you've solved it.

00:53:36.520 That's interesting.

00:53:37.220 Yeah.

00:53:37.400 I think if you make it totally transparent, it will be a barrier to the adoption of technology

00:53:43.680 that will be, on balance, immensely life-saving for everyone, you know, drivers and pedestrians

00:53:49.420 included.

00:53:50.600 We now have tens of thousands of people every year reliably being killed by cars.

00:53:55.640 We could bring that down by a factor of 10 or 100, and then the deaths that would remain

00:54:01.200 would still be these tragedies that we would have to think long and hard about whether the

00:54:05.060 algorithm performed the way we want it to.

00:54:07.120 But still, we have to adopt this technology as quickly as is feasible.

00:54:12.320 So I think transparency here could be a bad idea.

00:54:16.640 I think it's true.

00:54:17.220 I mean, I find that I know people have insisted they would never go into a self-driving car.

00:54:22.260 And I find this bizarre because the alternative is far more dangerous.

00:54:26.540 But you're right.

00:54:27.920 And I think there's also this fear of new technology where there'll be a reluctance to use them.

00:54:33.500 Apparently, there was a reluctance to use elevators that didn't have an elevator operator for a

00:54:38.140 long time.

00:54:38.820 So they have to have some schnooks stand there in order so people would feel calm enough to

00:54:42.740 go inside that.

00:54:44.180 That's hilarious.

00:54:44.740 But I agree with the general point, which is a more general one, which is there's no opting

00:54:50.380 out of moral choices.

00:54:52.720 Failing to make a moral choice over, say, giving to charity or what your car should do is itself

00:54:58.800 a moral choice and driven by a moral philosophy.

00:55:03.260 I also just can't resist adding, and I think this is from the Very Bad Wizards group, but

00:55:08.180 you can imagine a car that had a certain morality and then you got into it and it automatically

00:55:12.300 drove you to like Oxfam and refused to let you move until you gave them a lot of money.

00:55:17.640 So you don't want too moral a car.

00:55:19.920 You want a car sort of just moral enough to do your bidding, but not much more.

00:55:25.460 Have you been watching any of these shows or films that deal with AI, like Ex Machina

00:55:30.160 or Westworld or Humans?

00:55:31.640 I've been watching all of these shows that deal with AI.

00:55:35.460 And they all deal with, Ex Machina and Westworld all deal with the struggle we have when something

00:55:45.040 looks human enough, acts human enough, it is irresistible not to treat it as a person.

00:55:54.380 And there philosophers and psychologists and lay people might split.

00:55:58.620 They might say, look, if it looks like a person and talks like a person, then it has a consciousness

00:56:04.240 like a person.

00:56:05.040 Dan Dennett would most likely say that.

00:56:07.720 And it's interesting, different movies and different TV shows, and I actually think movies

00:56:11.580 and TVs are often instruments of some very good philosophy.

00:56:15.160 They go different ways on this.

00:56:17.180 So Ex Machina, I hate to spoil it, but so viewers should turn on the sound for the next

00:56:23.420 60 seconds, they don't want to hear it.

00:56:25.080 But there's a robot that's entirely, you feel tremendous empathy for her and caring for her.

00:56:29.980 The main character trusts her, and then she cold-bloodily betrays him, locks him in a room

00:56:36.660 to starve to death while she goes on her way. 0.74

00:56:39.260 And it becomes entirely clear that all of this was simply a mechanism that she used to win

00:56:45.740 over his love.

00:56:47.300 While for Westworld, it's more the opposite, where the hosts, I guess, Dolores and others

00:56:55.220 are seen as, they're really people, as viewers were supposed to see them as people.

00:57:00.380 And the guests who forget about this, who brutalize them, they're the monsters.

00:57:06.900 Yeah, it's very interesting.

00:57:08.480 I think all of these films and shows are worth watching.

00:57:12.100 I mean, they're all a little uneven from my point of view.

00:57:14.780 There are moments where you think, this isn't the greatest film or the greatest television

00:57:17.920 show you've ever seen.

00:57:18.660 But they all have their moments where they, as you said, they're really doing some very

00:57:24.580 powerful philosophy by forcing you to have this vicarious experience of being in the presence

00:57:31.300 of something that is passing the Turing test in a totally compelling way and not the way

00:57:38.100 that Turing originally set it up.

00:57:40.400 I mean, we're talking about robots that are no longer in the uncanny valley and looking weird.

00:57:46.300 They are looking like humans.

00:57:49.180 They're as human as human, and they are, in certain of these cases, much smarter than

00:57:53.980 people.

00:57:55.240 And this reveals a few things to me that are probably not surprising.

00:58:00.880 But again, it's to experience it vicariously, just hour by hour watching these things, is

00:58:05.640 different than just knowing it in the abstract.

00:58:08.240 That's right.

00:58:08.500 The best movies and films and movies and TV shows and books often take a philosophical

00:58:15.260 thought experiment, and they make it vivid in such a way you could really appreciate

00:58:20.380 it.

00:58:20.760 And I think that sentient, realistic humanoid AI is a perfect example of these shows confronting

00:58:27.200 us with, this is a possibility.

00:58:30.000 How will we react?

00:58:31.440 I think it tells us how we will react.

00:58:33.760 Once something looks like a human and talks like a human and demonstrates intelligence

00:58:41.360 that is at least at human level, I think for reasons I gave somewhere on this podcast

00:58:48.040 and elsewhere when I've talked about AI, I think human-level AI is a mirage.

00:58:54.260 I think the moment we have anything like human-level AI, we will have superhuman AI.

00:58:59.300 We're not going to make our AI that passes the Turing test less good at math than your

00:59:05.000 phone is.

00:59:06.280 It'll be a superhuman calculator.

00:59:07.940 It'll be superhuman in every way that it does anything that narrow AI does now.

00:59:12.580 So once this all gets knit together in a humanoid form that passes the Turing test and shows

00:59:18.780 general intelligence and looks as good as we look, which is to say it looks as much like

00:59:25.300 a locus of consciousness as we do, then I think a few things will happen very quickly.

00:59:30.800 One is that we will lose sight of the fact of whether or not it's philosophically or scientifically

00:59:36.860 interesting to wonder whether this thing is conscious.

00:59:39.740 I think some people like me, you know, who are convinced that the hard problem of consciousness

00:59:44.480 is real might hold on to it for a while.

00:59:47.400 But every intuition we have of something being conscious, every intuition we have that other

00:59:52.740 people are conscious, will be driven hard in the presence of these artifacts.

00:59:59.080 And it will be true to say that we won't know whether they're conscious unless we understand

01:00:03.940 how consciousness emerges from the physical world.

01:00:07.200 But we will follow Dan Dennett in feeling that it's no longer an interesting question because

01:00:13.560 we find we actually can't stay interested in it in the presence of machines that are functioning

01:00:18.720 at least as well, if not better than we are, and will almost certainly be designed to talk

01:00:25.860 about their experience in ways that suggest that they're having an experience.

01:00:30.200 And so that's one part that we will feel, we will grant them consciousness by default, even

01:00:35.740 though we may have no deep reason to believe that they're conscious.

01:00:38.500 And the other thing that is brought up by Westworld to a unique degree, I guess humans also, is

01:00:46.360 that many of the ways in which people imagine using robots of this sort, we would use them

01:00:51.800 in ways we at least we imagine that we wouldn't use other human beings on the assumption that

01:00:55.900 they're not conscious, right?

01:00:56.960 That they're just computers that really can't suffer.

01:00:59.660 But I think this is the other side of this coin.

01:01:02.320 Once we helplessly attribute states of consciousness to these machines, it will be damaging to our

01:01:08.460 own sense of ourselves to treat them badly.

01:01:12.800 We're going to be in the presence of digital slaves, and just how well do you need to treat 1.00

01:01:17.660 your slaves? 0.70

01:01:18.420 And what does it mean to have a super humanly intelligent slave?

01:01:22.440 I mean, that just becomes a safety problem.

01:01:24.380 How do you maintain a master-servant relationship to something that's smarter than you are and

01:01:28.960 getting smarter all the time?

01:01:30.620 But part of what Westworld brings up is that you are destroying human consciousness by letting 0.98

01:01:37.480 yourself act out all of your baser impulses on robots, on the assumption that they can't

01:01:43.100 suffer, because the acting out is part of the problem.

01:01:46.820 It actually diminishes your own moral worth, whether or not these robots are conscious.

01:01:52.720 Right.

01:01:52.880 So you have these two things in tension.

01:01:54.680 One is that when it starts to look like a person and talk, it'll be irresistible to see

01:01:59.740 it as conscious.

01:02:00.420 You know, you could walk around and you could talk to me and doubt that I'm conscious, and

01:02:04.840 we could doubt that about other people.

01:02:06.300 But it's an intellectual exercise.

01:02:08.000 It's irresistible to treat other people as having feelings, emotions, consciousness, and

01:02:15.600 it'll be irresistible to treat these machines as well.

01:02:18.080 And then we want to use them.

01:02:20.240 And so in Westworld is a particularly dramatic example of this, where characters are meant to

01:02:25.700 be raped and assaulted and shot, and it's supposed to be, you know, fun and games.

01:02:31.900 But the reality of it is these two things are in tension. 0.91

01:02:35.340 Anybody who were to assault the character Dolores, the young woman who's a robot, would be seen

01:02:44.900 as morally indistinguishable from someone who would assault any person.

01:02:49.260 And so we are at risk for the first time in human civilization of, in some sense, building

01:02:56.740 machines that we are in morally, it's morally repugnant to use in the sense that they're

01:03:02.740 constructed for.

01:03:03.860 Yeah.

01:03:04.080 It would be like genetically engineering a race of people, but wiring up their brains so 0.92

01:03:09.680 that they're utterly subservient and enjoy performing at our will. 0.86

01:03:13.740 Well, that's kind of gross.

01:03:15.220 And, and I think we would, we're very quickly going to reach a point where we'll see the

01:03:20.660 same thing with our machines.

01:03:22.880 And, and then what I would imagine is, and this goes back to building machines without

01:03:26.700 empathy or perhaps without compassion is there may be a business in building machines to do

01:03:32.740 things that aren't that smart.

01:03:34.720 I'd rather have my floor vacuumed by a Roomba than by somebody who has an IQ of 140, but is 0.82

01:03:41.460 wired up to be a slave.

01:03:42.440 I think the, the humanoid component here is the main variable.

01:03:47.960 If it looks like a Roomba, you know, it doesn't, it actually doesn't matter how smart it is.

01:03:52.020 You won't feel that you're enslaving a conscious creature.

01:03:56.300 What if it could talk?

01:03:57.620 It comes down to the interface.

01:03:59.180 And so far as you humanize the interface, you drive the intuitions that now you're in

01:04:04.080 relationship to a person.

01:04:05.880 But if you make it look like a Roomba and sound like a Roomba, it doesn't really matter

01:04:11.240 what its capacities are as long as it still seems mechanical.

01:04:16.600 I mean, the interesting wrinkle there, of course, is that ethically speaking, what really

01:04:21.460 should matter is what's true on the side of the Roomba, right?

01:04:25.000 So if the Roomba can suffer, if you've built a mechanical slave that you can't possibly empathize

01:04:32.020 with because it doesn't have any of the user interface components that would allow you to

01:04:37.100 do it, but it's actually having an experience of the world that is vastly deeper and richer

01:04:44.560 and more poignant than your own, right?

01:04:47.800 Well, then you have just, the term of jargon now in the AI community, I think this is probably

01:04:52.980 due to Nick Bostrom's book, but maybe he got this from somewhere, the term is mind crime.

01:04:58.200 You're creating minds that can suffer, whether in simulation or in individual, you know, robots.

01:05:05.120 This would be an unimaginably bad thing to do.

01:05:08.840 I mean, you would be on par with Yahweh, you know, creating a hell and populating it. 0.99

01:05:14.560 If there's more evil to be found in the universe than that, I don't know where to look for it.

01:05:19.840 But that's something we're in danger of doing insofar as we're rolling the dice with some

01:05:25.580 form of information processing being the basis of consciousness.

01:05:29.360 If consciousness is just some version of information processing, well, then if we begin to do that

01:05:35.240 well enough, it won't matter whether we can tell from the outside.

01:05:38.620 We may just create it inside something we can't feel compassion for.

01:05:43.440 That's right.

01:05:43.980 So there are two points.

01:05:44.980 One point is your moral one, which is whether or not we know it, we may be doing terrible

01:05:51.420 moral acts.

01:05:52.180 We may be constructing conscious creatures and then tormenting them.

01:05:55.700 Or alternatively, we may be creating creatures that are machines that do our bidding and have

01:06:01.720 no consciousness at all.

01:06:02.860 Well, it's no worse to assault the robot in Westworld than it is to, you know, to bang a

01:06:09.920 hammer against your toaster.

01:06:11.600 But so that's the moral question.

01:06:13.620 But it still could diminish you as a person to treat her like a toaster.

01:06:17.540 Yes.

01:06:18.200 Given what she looks like. 0.97

01:06:19.380 And that's I mean, so so raping Dolores on some level turns you into a rapist, whether 0.99

01:06:24.360 or not she's more like a woman or more like a toaster. 1.00

01:06:27.360 Yes.

01:06:27.880 So so this is akin to this treatment of robots is akin to I forget the philosopher.

01:06:32.780 It may I forget who the philosopher was.

01:06:36.040 But the claim was that animals have no moral status at all.

01:06:40.500 However, you shouldn't torment animals because it will make you a bad person with regard to

01:06:45.300 other people and people count.

01:06:47.480 And and it's true.

01:06:48.340 It's it's I mean, you one wonders one after all, we do all sorts of killing and harming

01:06:56.320 of virtual characters on video games.

01:06:59.520 And that doesn't seem to transfer.

01:07:01.640 It hasn't made us worse people.

01:07:03.380 If there is an effect on increasing our our our violence towards real humans, it hasn't

01:07:08.700 shown up in any of the homicide statistics or and the studies are a mess.

01:07:12.860 But I would agree with you that there's a world of difference between sitting on my Xbox

01:07:17.380 and shooting, you know, aliens as opposed to the real physical feeling, say, of strangling

01:07:23.520 someone who's indistinguishable from a person.

01:07:26.240 And and that's the second point, which is even if they aren't conscious, even if as a matter

01:07:31.820 of fact, from from a God's eye view, they're just things, it will seem to us as if they're

01:07:39.520 conscious.

01:07:40.000 And then the act of tormenting conscious people will either be repugnant to us or if

01:07:46.600 it isn't, it will lead us to be to be worse moral beings.

01:07:51.200 So those are the dilemmas we're going to run into probably within our lifetimes.

01:07:55.400 Yeah, actually, there's somebody coming on the podcast in probably a month who can answer

01:07:59.580 this question.

01:08:00.060 But I don't know which is closer.

01:08:01.960 Realistically, machine intelligence that passes the Turing test or robot interface, you know,

01:08:09.200 robot faces that are no longer uncanny to us.

01:08:14.340 Yeah.

01:08:14.620 I don't know which will be built first, but it is interesting to consider that the association

01:08:19.280 here is a strange one.

01:08:22.160 If you'd like to continue listening to this conversation, you'll need to subscribe at

01:08:28.260 samharris.org.

01:08:29.660 Once you do, you'll get access to all full-length episodes of the Making Sense podcast, along

01:08:34.160 with other subscriber-only content, including bonus episodes and AMAs and the conversations

01:08:39.680 I've been having on the Waking Up app.

01:08:41.800 The Making Sense podcast is ad-free and relies entirely on listener support.

01:08:45.880 And you can subscribe now at samharris.org.

01:08:58.260 Thank you.

Making Sense - Sam Harris - December 12, 2016

#56 — Abusing Dolores

Episode Stats

Summary

Transcript