How Data Reveals Everything About Trust, Teamwork, and Power — Alex Pentland

0:00What you're showing is that you want to know the person as a person. They're not just, you know, a facility for work. A commodity. Yeah.

0:14What would happen if we just told the truth? Welcome to TruthWorks, where we dig into the nitty gritty of leadership and work. And what needs to change. I'm Jessica Neal. And I'm Patty McCord. Our journey together started in HR. But trust us, it's evolved into something wild, honest, and well, a bit rebellious. So throw out the handbook. We're here to redefine rules to work for us, not against us. Let's dive into another episode.

0:51Hi, everyone. I'm Jessica Neal. Welcome to another episode of TruthWorks. Today, I have this person on the podcast who I'm excited to talk to you guys about. You're famous in your own right here, but this person's name is Alex Sandy Pentland. And you are, I don't even know how to explain you. You're like a distinguished award-winning data scientist, right? And you have won many awards, but I think your full-time job is that you're a professor at MIT and you're behind sort of this whole lab that I want you to tell us about that you have lots of, I'm guessing, experiments going on in. But tell us a little bit about you, Sandy.

1:38I know you like to go by Sandy, not Alex. So tell us a little bit about you and the work that you're doing. And then I'm going to get into some of the questions that I have for you. Okay. So I'm Sandy. My dad's name was Alex, so I got the diminutive, Sandy. So I'm at Stanford now. Although I thought you were still at MIT. Well, I am sort of still at MIT, right? So at MIT, it's the Media Lab, which is, what is Media Lab? Well, it's actually sort of about how humans and technology work together. And at Stanford, it's the Human AI Initiative, which is how human and AI work together. So that's really my core focus is, yeah, sure, data, but it's not like statistics.

2:22It's about how do you understand what people are doing using data to be able to really get past some of the myths and some of the advertising about how life and cities and companies and intergenerational things happen. And I think that what we're trying to do is take it to a new level where we have a lot more truth about human life and what's good and what's maybe not so good and where we're going. Yeah, which is what I'm so fascinated to talk to you about today. And, you know, I guess so I think a good way to categorize the work that you do, Sandy, is that you're studying the human behavior, invisible systems that like shape how we work, how we make decisions, how we figure out how to trust each other.

3:10Is that a fair categorization of your work? Yeah, absolutely. Yeah. Yeah. So I think what I really want to explore today is that beneath all the data and the models, what does it actually mean to like be human, right? Within this world of AI, in this world to data. How do you get trust? How do you collaborate? How do you lead when there's a lot of our behavior that's being like measured and, you know, like metrics. So I guess the first question I really have for you is about trust and its currency. I think you've talked about that a lot. So can you expand on like what you've researched and seen in a high trust environment, not with necessarily leaderships and systems, but just how have you seen that play out?

4:07What's the research that you've studied? Tell us a little bit more. Do you think we've lost our way? Lots of, well, I think we misunderstand what trust is. And there's sort of too much of, you know, people trying to manage trust, right? I mean, look, trust is in our data. We've looked at lots of different situations in companies and communities and so forth. It's really when two people, because it's between people, right? When people feel like the other person has their back. So, you know, every time I talk to you, I get something good from it. You're obviously putting effort into it. And every time, you know, you talk to me, I try to do the same.

4:50People are very good at sensing that. And what happens is, of course, that people end up going to each other for help, for advice, for moral support, for fun. And when you see that behavior of people, you know, choosing to seek out the other person and vice versa, that's a very strong indication of trust. And if the situation is right, it's also a precursor to friendship. I mean, when I say the situation is right, you know, at work, sometimes things are a little formal or you don't see people very often. But, you know, we found that we can actually predict when people are likely to become friends because they're beginning to have that exchange with each other.

5:31So trust really is this sort of notion that, you know, we're valuable to each other and we are happy with the deal essentially that we have by interacting with each other. It's not some necessarily big thing. Like for instance, you know, if you were buying something at the local, you know, corner store, right? And you went there every day, maybe you bought a coffee on the way. And one day you show up without your wallet and you can't pay. Would they just give you the coffee anyhow, because they know who you are, right? That's the sort of genius of trust is that there's a little bit of helping people out in there.

6:14And people forget it, right? That's what the core is. Right. And at work, because we like to talk a lot about work on this show, In your research and studies, how many companies are maybe not exceeding their potential because they've just lost trust? Well, oh, you know, 95%. Yeah. It's everywhere, right? Right. And it's because they think that they confuse trust with being predictable, right? Right? So, you know, we have rules, we follow them, you can trust us, right? Well, no, I can trust you to exploit me or ignore me. But trust is really, you know, having that level of interaction. Culture, like the culture of trust, has to do with people talking to each other, trading stories about what happened, about what the stories mean.

7:15And if the stories are about how good things happened when there was a problem, and that type of story is spreading around, then you get a culture of trust because people have an expectation that things will come out okay and they'll be understood and so forth. It's that sort of circulation of stories that builds the culture, builds this, what I'm calling in my new book, shared wisdom, that is the stuff of trust. No, it's so true. And I mean that, so at Netflix, we worked really, really, really hard to have trust because we knew that if we had that connection from the top down, that we would move faster.

8:03We would make better decisions. We'd work better with each other when we didn't agree. We would work through it faster, right? But without the foundation of trust, we knew that things wouldn't be as easy. And so we worked really hard to push each other to have the conversations nobody wanted to have to talk about the things that nobody wanted to talk about. And really get to know each other on a, not just work level, but a personal level. And I'm not talking like, you know, I think we were colleagues more than friends, but we knew about each other's worries and strengths and hardships, right? Yep, yep.

8:51It's like whenever I do a Zoom call, I always spend the first couple minutes asking how people are and how's the family. I mean, just dumb things like what's the weather like because that gets into all sorts of things. what you're showing is that you want to know the person as a person. They're not just a facility for work. Right. A commodity. Yes. Yeah. So how do you... Okay, I want to talk about the new book, right? But I also want to talk about how do you do your research? What's the process? Tell us where you look, what you've studied. How have you learned all these wonderful things that you know?

9:31Well, we do a number of different things. the overall area is called computational social science. So everyone's sort of familiar with social science, and it actually has gotten a bit of a bad reputation because a lot of things turn out to be not replicable or it's not clear how you use them in the real world. Computational stuff is saying, oh my gosh, we have all this data out there now about where people go and what they say online and this, that, and the other thing. and that's really scary, but if you approach people and say, look, what I'd like to do is not look at your personal lives, not steal your data, but what I'd like to do is understand what are the signals of trust, what are the signals of interaction that mean that your kids will grow up to be successful, things that they care about, and what we're asking them to do is to contribute a little bit of data so that everybody can understand what's a better way to live.

10:31So you find things, like I talked already about how it's this pattern of interaction that is the signature of trust, how it's the spread of ideas and conversations among a group of people that is how you build culture. There's other things that are really surprising, not my work, but from my community. like if you take kids that are at risk, right? So they, you know, perhaps a bad family situation, bad neighborhood, but there's at least one person in the neighborhood that they get to see that has been successful. They've started a store, they've done something, they've sort of taken charge of their life.

11:11Those kids are so much more likely to grow up to be successful themselves. It only takes one. Interestingly, if you're a girl, it has to be a woman that's successful. They don't pay attention to them. If you're a boy, it has to be a man that was successful. And it doesn't matter that there's some women that are successful. So it's this sort of trading of stories. And in this case, these are stories about how you have to act to be successful in your life. That is what gives people sort of a repertoire of behaviors to choose from. And it's that repertoire of ideas. What do you do in this? What do you do?

11:53That's sort of the wisdom that can guide you through life. And it's also shared wisdom like that, sort of things that everybody knows is how you get people to work together. Mm-hmm. That's, well, that's just fascinating. And I think your work is needed now more than ever. and like the world that we're headed into. But tell us a little bit about the new book. What's in the book? What are you writing about? Well, you know, it's this topic, really. It's this notion that we don't really understand what it means to trust, to have culture. The title of the book is Shared Wisdom, which is the sort of things that we all as a community believe.

12:37And the subtitle is Cultural Evolution in the Age of AI. And what that is, is saying, where are there opportunities for all this AI stuff? It's coming. You got to like figure out what you're going to do with it, right? Yes. Where are there opportunities to build a better culture? So what's better? Well, better is more trust, more collective action to be able to make your lives and communities better, making better decisions. And it turns out that the rules for doing this are not that different than the rules were 100 ,000 years ago, right? People have to talk to each other. You have to be able to sample, talk to people with lots of different experiences to figure out what works, what doesn't work.

13:21So what the AI that's going to be useful for us as people is AI that helps us find the stories that are most useful to us and to be able to communicate with each other. But don't you think that, I don't know, I feel like AI too, like you just said, that it's important that we get like different perspectives, different stories, like get all all the information. But I feel like sometimes with AI, you're only getting that story that it wants to promote to you because of one thing that you like. So you're getting this, you know, you're getting insulated with the same type of people, the same type of perspective.

14:03And it's almost, I don't know, I feel like these days more harmful than good. Like you think about how Instagram works. And I personally feel, and I want to know your thoughts because you're way smarter than me, but I feel like it's making us less connected and more divided. I absolutely agree. But you've made a move in there, which I think is the critical move. And that is, you're assuming that AI is to provide answers. Okay, well, I don't want AI to provide answers. I want AI to connect people. I want AI to be able to be a fair mediator among people during a conversation. And I'll explain what I mean about that.

14:45So let me give you an example of something that companies are doing, which is really interesting. I was on a panel with the chief technical officers of a bunch of big companies, and they're all talking about how it's really hard to get data to do things and this. But one of the things they had all done is build AI buddies. So what's an AI buddy? An AI buddy is all the manuals that nobody ever reads. It's all the newsletters. It's how do you find people who are doing things like you or relevant to what you're doing? Those are the questions that it answers. It's trying to connect you to the right people for what you're doing right at the moment.

15:25That is a very human thing. And it's something we have a hard time and even increasingly hard time because of work from home, because of international. So, you know, AI as a buddy that helps you find the things doesn't tell you what to do. It says, here's some things that might be interesting. It's like Google search. Google search is not unhuman because people use it to help them find things out. It's like a super powered library index card or something like that. AI is the same, can be used the same way. And then the other thing we do, we built a website. We're doing this with cities and companies and so forth, all free, all open, just take it, called deliberation.io.

16:10And what it does is it's AI helping to be a good facilitator. So the AI doesn't contribute any ideas at all. It says, it listens to the conversation and it says, occasionally, what I hear is some people saying this and other people saying that. Is that right? And that sort of brings people back to what has actually been said. So the loud voices tend not to dominate as much. So the minority voices don't get lost. And here's the kicker. The decisions that people make when they have that type of mediation are objectively considerably better than the decisions they make without mediation. And people like it better, right?

16:57How do you get it to mediate? How do you utilize that in a mediate for people who want to do that? Well, we have another something where it's like, think about it like a Slack channel or Twitter or something like that. People are making comments. They don't get to reply to each other. The comments are anonymous, but you know they're from the work group. Okay. So you don't know quite who's saying it. Okay. And what it does every once in a while, it gets the instruction to please summarize what people are saying. That's all it does. Okay. It's not perfect at that. Right. Gotcha. We understand. But it's not bad.

17:34It actually seems to be better than the normal human in terms of bias and forgetting. Right. Just summarizing, not finding the answer. So by summarizing, what you're doing is reflecting back to people what has actually been said. You're picking up the threads that might otherwise get lost. And what ends up happening is people begin to understand each other better. Let me give you an example of this. We did this with a couple of different things. The biggest one, we had a group of 30 ,000 people, right? And we were debating things like, you know, who should have guns? Really contentious things.

18:13Yeah. And what we found was that the biggest problem in that sort of debate is that most people don't know what the other side thinks because all you hear are these like influencers and the big voices like the rhetoric the rhetoric and so you think everybody on the other side is a crazy yeah and yes there are crazy can we give this mediator to our government right now well interesting you say that so we did a exercise with washington dc city. And we asked a bunch of people and we tried to make this very inclusive. So people with two jobs and three kids, that sort of people, right? We're not talking about the big brain, powerful people.

19:00We're talking about the people. Every day, the everyday people. And we asked them, what should government do with AI? And you know what they came up with? I got to have you guess. To have them less divided. No, they wanted an AI to help them deal with the government. They say the government is too crazy. It's too complicated. God knows what I'm going to have to do. That's the sort of thing that AIs are good at. It's like search. It's like collecting the data to give to you so you can make your decisions in an informed way where you know the context. And they like that idea. Everybody was using this moderation AI.

19:43I don't think people, we told them, we showed them, But I don't think anybody like sort of fascinated on the fact that, oh, hey, here I am, interacting with an AI and all these other people. And it feels natural. They like the experience. And they came up with something that the government didn't think of. The researchers didn't think of. It was, you know, we need AI for us to defend us. Yeah. Wow, that's so fascinating. So, you know, as I'm listening to you talk, I think what I'm hearing is if you're using AI in your life, in your company, from a connection lens, from a summary lens, from a listening lens, instead of like, tell me what to think lens, that can be very beneficial.

20:36Right. Because people, that's exactly correct. And the reason is, you know, people, it's hard to talk to a group of 100 people, right? Yeah. You just physically can't do it. But you can have somebody summarize 100 people. Often the way it ends up happening, though, is somebody who has an agenda is the one doing the summary. Mm-hmm, right. So actually, it's pretty good because it might be a little biased, might be a little this or that, but it doesn't have an agenda. Right. And you can actually make sure it doesn't have an agenda. You can tell it not to. Yeah, exactly. You tell it not to. And so it's like holding the community up to a mirror, right?

21:19Yeah. Oh, look. Right? Yeah. I have a friend that... Sorry, I have a friend that uses this type of technique to run a government in Taiwan. Wow. He calls it the face of the crowd. When people see the face of the crowd, they behave very differently than they normally do because now they know the other people better. And that's what this is doing. This is giving people this mirror so that they can see themselves better. Yeah, so how does he do that in Taiwan? What does he do? What is he doing exactly? Well, the things that we're doing are inspired by what Xi actually is doing. Xi? I know, here I said he, and I should have said, I should have said, what are they doing?

22:02I just assumed it was a man. Yeah, no, no. Easy mistake. Actually, it's a little difficult because when I first met Audrey, he was a man, and now he's not. Okay, okay, well, there you go. Which is pretty interesting for somebody who's a state official, right? Yeah, fascinating. Anyway, so what they did is this system called POLIS, which was actually developed in San Francisco. And we took the basic ideas from that and tried to make them much more inclusive and effective. And the basic ideas are that people shouldn't be having like little battles the way they do on Twitter or something like that.

22:44You can make comments and you need to see the other people's comments. So there's mechanisms to make sure you see other comments, but not like, you idiot, you can't do that, right? All the people are authorized in some sense. You know that they're actually humans. Right. They're not robots. It's not some robot. Yeah. But you don't know their names. They're still anonymous. So it's like secret voting and stuff like that, secret ballot. And then there's visualization of what people are saying so that you can actually see the face of the crowd. And the way this original system that they were using did it was sort of complicated and people didn't quite get it.

23:21So we made it really something that is like what you see on your phone. In fact, it runs on your phone, right? You can sort of begin to engage this, we call it everyday deliberation. You know, people talk about stuff all the time. What do you think of this? What do you think of that, right? But it's not like you're studying it. It's just this little conversation around the coffee pot or before the meeting or whatnot. And that's what we tried to do is sort of have people put in comments and then be able to visualize what the whole range of comments is so that you're not dominated by the loud voices and the you know the the influencers and stuff like that so you see what everybody really thinks and it turns out that when you see that you realize that you're not alone you're not the outlier then the other guys are not crazy mostly and so maybe a little right maybe well some are I guarantee you, right?

24:17Yes. There are some on both sides that are a little crazy, you know. Absolutely. There's a spectrum always. But like even, for instance, just to give an example, it's a very salient type of example. You know, in this country, an issue like abortion is very, very contentious. But actually, if you ask individuals in sort of an anonymous way, most people are in agreement. and you know the agreement in the united states is a little more liberal than the agreement in europe but it's not that different right but you never hear that because there's a lot of passion by people with very loud voices right and i think you know there's a lot of issues like that yeah where i think that's so fascinating that that could you know if we just used ai a little differently than we're using it now to help us solve some of these tougher problems or at least get on the same page about really how people are actually feeling versus you know the extreme end on both ends like that would be really productive that would be that's just it would be great can we just start doing that that would be wonderful let's do let's start doing that tomorrow but if i'm we're trying to do that right yes well sandy i hope you get it done If I'm a CEO and I have a company and, you know, we're trying to find our way in AI and perhaps I'm trying to build more trust, perhaps I'm trying to have more transparency.

25:52What's the one or two things you think I should do, you know, immediately in terms of helping people with this? That's amazingly effective or seems to be is this AI buddy thing. So don't use one of the big expensive AI models. Use the small ones that are open source, but put into it all the manuals and newsletters and stuff that are relevant for each job that people have. To have something that can help people know what's going on in this company that they're working for. What are people really doing? And that's the sort of face of the crowd. In this case, it's the face of the company, as opposed to people who are blustering or angry or something like that.

26:41You'd like to find out what's really going. So you can find the people that you really want to talk to, the people that have things to contribute to what you're doing. So you know what the culture is. I think that's maybe the simplest thing to do and in some ways the most effective. It helps a lot with work from home. It helps a lot with distributed sites around the world. it helps people be in the loop. I mean, this is not rocket science. Yeah. This is like, you know, humans need to be in the loop. And so a tool that helps you be in the loop makes a lot of sense. And then the other thing, you know, I think is really easy is this deliberation I.O.

27:19stuff that is just a simple little thing. It runs on phones, runs on computers. It's free. It's all there. It's just a way to have conversations that are civil where you get to hear all the voices. and it turns out people are civil and you can hear all the voices, they make better decisions. Duh, right. Of course they do. But, you know, that's the thing to do. And the reason not to use AI is that AI is trained on historical data, always. Right, yeah. It has to be, right? And it's telling you what to think, but you don't know why. If you learn what people are, this sort of like shared wisdom, what people think in general.

28:01What happens is people begin to understand why other people think this way and why this, some particular thing is a good solution for everybody. And that's what you need. That's right. You need that sort of shared sense of what we're doing, why, in order to be able to move forward together. Yeah, because your human intuition, right is just being you know you're not not using that you're just informing it with more information because you're an imperfect human and maybe you forgot what happened in the meeting or somebody didn't summarize it correctly so you're using ai to help you hear all the things you need to hear so that that human intuition is more fully formed if you will that's right That's right.

28:51Yeah. And I mean, you know, the human intuition is not magic. It's something that's much more informed by context and experience than AIs. I mean, AIs never lived, right? They never had a mother. They never had a father. They never had a bully in middle school. There's a bunch of things that they never really had. Yeah. they they are sensible but they're sort of the common conversations public conversation and that's not that's different than the way we actually live through things and uh the internal conversations we have yeah ourselves yeah and so getting rid of those and just ignoring that internal experience and and the context is is a mistake yeah yeah interestingly in the book i point out many examples where, you know, it looked like a good idea because this was what all the scientists or politicians or whatever said, but actually it was stupid.

29:52Tell us one of those. I want to know one of those. Tell us one of those that you talk about in the book. Well, you know, there's a number of different things. Different people find different things striking, right? So, I mean, this is one you may not relate to. We'll do a different one. So, you know, like you get on TV and you hear all these physicists talking about the Big Bang and the beginning of the Earth and how they're like so authoritative. They really, I was like, my God, this guy, this guy's like God. He knows like how the world, the universe began, right? Except that about 20 years ago, they discovered this stuff called dark energy and dark matter.