What AI can’t replace—and why it matters for value

0:00The data are very clear at this point that when people feel psychologically safe, they innovate more.

0:06Jamil Zaki: Same goes for empathy. You feel that, hey, my leaders have my back. Well, then you get to take more risks. You get to be more creative. When you look at companies that are producing real AI maturity, it's because they're asking their employees to be their best selves using this technology as opposed to threatening them silently or otherwise with replacement. move.

0:28Bryan Hancock: That's Jamil Zaki. He's a professor at Stanford University, and he joins Brooke, Brian, and me to talk about empathy in the context of AI, specifically how leaders who routinely demonstrate empathy are more successful in driving widespread adoption. This is McKinsey Talks Talent. I'm your host, Lucia Raheli, with talent experts Brooke Weddle and Brian Hancock.

0:58Bryan Hancock: Professor Jamil Zaki, welcome to the podcast.

1:01Jamil Zaki: It's my pleasure to be here.

1:02Bryan Hancock: It's so great to see you again. And Brooke and Brian, welcome back. It's great to be here. It's great to be here. So, Jamil, it seems to me that empathy is having a moment. Why do you think empathy is having its star turn?

1:19Jamil Zaki: Well, you know, I've been studying this topic for almost 25 years now, and I can tell you that empathy has had many moments over that time, some in which people seem to love it and others in which people seem to really not love it at all I think that one reason that it's having a moment in this decade is because we are lonelier you know a glass of water might be the least interesting beverage on the planet but if you've been hiking through the desert for 12 hours it's the most interesting thing on earth and I think that since the pandemic many of us have felt an increasing sense of isolation and disconnection And when you feel that, it's like an alarm system goes off in your mind and you ask yourself, how can I find a greater degree of connection?

2:02Jamil Zaki: And empathy is, of course, one powerful tool to do just that.

2:06Bryan Hancock: And just revisiting our previous conversation briefly, talk to us about the value of empathy and human connection, which you say is diminishing.

2:17Jamil Zaki: Yeah, I mean, I think that in the workplace, for instance, we can make two really powerful cases for empathy. The first is a human case. Most adults spend most of their waking hours at the office. And right now, many of them are not enjoying that at all. Engagement at work has plummeted in the last several years and connection has as well. So I think that the human case for empathy at work is that when leaders and colleagues connect with us, when they make us feel understood and cared for, we, employees, feel better. Our happiness rises, mental health improves. People who feel that their colleagues are empathic actually call in sick less often with stress-related illness.

2:58Jamil Zaki: So that's the human case, is that we can be better for each other in workplaces. There's also a business case, which is that when people feel connected, they also do a lot better at work. I mean, I think disengagement costs the global economy like$9.8 trillion a year. And so anything that we can do to make people feel more connected to their work is really powerful economically. When people feel cared for by their leaders, they're also more likely to take creative risks with their work, to innovate, to share information and skills with their colleagues, and to collaborate more efficiently. So I think that there's just multiple levels at which connection at work, although far too scarce, is really powerful when we can build it.

3:39Jamil, one thing that I've noticed in my work with organizations these days is just the words anxiety come up a lot. So mostly related to AI, right? So many people feel anxious, nervous around what AI and how AI will impact them and their jobs. As you see the role of social capital, which you mentioned playing out, are there certain groups that you see impacted more? So we did research kind of coming out of the pandemic that showed that social capital was lower, especially for women and especially for junior level employees. Are you noticing any trends with respect to how that's playing out across different groups now?

4:23Jamil Zaki: Absolutely. And I think that the second case that you mentioned just now is the one that I see the most, which is younger employees. people who are trying to enter the workforce or rise from their first position to a second one. These are the folks who are really feeling quite unsafe, I would say now, because AI, for instance, is generally replacing sort of those more junior roles much more quickly than it's replacing senior roles. So you have young people who feel that really their ability to have the career that they thought they were going to have, not the career they thought they were going to have 10 years ago, the career they thought they were going to have two years ago.

5:01Jamil Zaki: seems to be disappearing from before their eyes. And this is creating an enormous amount of, well, psychological unsafety. Anxiety, as you said, there was a recent study that found through a sort of longitudinal analysis that after companies have big AI rollouts, their employees, especially their junior employees, become more depressed because of a lack of psychological safety. And through this, I think you also get an enormous backlash among those same folks who now don't want to participate in the revolution that is replacing their jobs and their opportunities. Yeah, absolutely. It's almost as if some of those younger employees have been hit twice, once by the pandemic, where they, you know, lack some of that connection, and now again with AI.

5:45So it's hard to get on solid footing from, you know, developing and kind of, you know, advancing in your career after those two shocks, right, that are playing out. And this is also where I'm seeing in my clients, they're trying to figure out what to do with it because they're seeing on the one hand, the entry-level employees are no more engaged, if anything, less engaged than they've been before, but turnover is also down. And so when employers are starting to think about that, one of the things that I've been talking them through is, okay, well, how do you lean into your managers? How do you help them be more empathetic, more present, more there versus maybe some other interventions or levers that you have?

6:26But it comes back to some of that human-to-human connection and reinforcing that with the entry-level folks so that they can feel comfortable and in some cases comfortable recognizing that, hey, there may be a different path that makes the most sense.

6:40Jamil Zaki: That's a really great point, Brian. And it's also very interesting that really what you have is low turnover and low engagement. Typically, those are two things that are anti-correlated with each other. But now you have people who are bringing their bodies to work, but maybe not all of their mind. And I think you make a great point that this is, in fact, the exact time to give empathy a bigger moment. I don't think that's necessarily intuitive, that this technological change should necessitate more empathy. But I think it really does. I mean, middle managers are really the ambassadors of an organization's culture for frontline employees, for these younger, more junior folks who we're talking about.

7:19Jamil Zaki: And I think that right now is the time to bring empathy training and upskilling on human connection to that layer of organizations in particular. I think that this is especially important because the things that managers used to do, you know, the knowledge that they have, for instance, junior employees can now go to AI to get that knowledge. So you have to ask yourself, what about a manager is least replaceable in this moment? And to me, it's the idea of having somebody, a human being who cares about you, who knows you and who is on your side. Those are the exact skills that will be most important and more important now during the AI age than they were before.

8:01Jamil Zaki: There was a report that came out from Workday on human connection in the workplace now. And it found that, you know, first of all, people are using AI. They feel a lot more efficient, more confident in their role. But they found that the predominant use case of AI, 76 % of employees that they used it for this, was not to make work tasks more efficient, but to get advice, including advice on how do I interact with my colleagues. And that coincided with people turning to Gen AI to basically give them advice and mentorship, frankly, on their work coincided with them feeling less comfortable turning to their colleagues and human mentors for the same advice.

8:40Which we've seen play out with coaching as well, right? So I am more comfortable being coached by a bot because the bot doesn't judge me, right? And so I think it's really interesting. And I love, you wrote a fascinating article about this, what can we learn about human connection and empathy from AI, which was a fascinating premise. And it would tell us a little bit more about that. But I mean, my thinking was it can only go so far, right? There are things that we can learn. You talked about the listening and playing back and some of the techniques that AI uses. But at the end of the day, there were some shortcomings that actually humans can uniquely play a role for.

9:19Jamil Zaki: Oh, yeah. I mean, it's astonishing. When these technologies first exploded onto the scene, people in my corner of the world who study empathy and connection started running all of these studies to see how empathic people perceive AI to be. And the predominant model here is what you could call like an emotional Turing test of sorts. So you write about something that's bothering you and you receive support from two different chats in return. You don't know which one came from a human stranger and which one came from a bot. And you rate how empathic each one is, how supported you feel, and how much better you feel after reading each of these responses.

9:58Jamil Zaki: My goodness, were we surprised when we started seeing these results. Chatbots simply ran circles around human strangers when it came to presenting, providing empathic and supportive responses. I mean, much stronger in terms of their ratings. And it wasn't just that they outpaced human strangers. They did better than counselors. They did better than physicians. And it was really, you know, so to your question, absolutely, Gen.ai appears to be empathic and it appears to be quite empathic. And why shouldn't it? It has nothing but time and energy for you, right? Its responses are very long and it's trained on the entirety of therapy that's ever been recorded, right?

10:41Jamil Zaki: So it understands techniques such as identifying your emotions, repeating back to you what you said and validating your experience. Whereas us human beings, you know, we are imperfect creatures. And when people come to us with their problems, we tend to jump to giving them advice, trying to solve their problems, or worse, talking about ourselves. So AI avoids those pitfalls, to your point. And that's why it's often viewed as more empathic. But as you also said, there's a paradox, because when you unveil that this response came from a chatbot people do not like to have been supported by a machine.

11:20Jamil Zaki: They much prefer empathic responses that come from a person. And I think that's for a reason. Empathy isn't just about saying the right things. It's about experiencing emotion alongside someone and genuinely caring for their well-being, wanting for them to feel better. AI can maybe look like it understands us, but it can't feel anything, at least as of now. and it doesn't genuinely care about us. And I think people can tell that. And over the long term, receiving support from chatbots, even really high quality support, leaves people lonelier than they were before.

11:54Bryan Hancock: Jamil, I so want to believe that. But how do you dovetail that with the rise of synthetic relationships?

12:05Jamil Zaki: Well, I think it's actually not that hard to square that circle. I think that there's abundant evidence now that connecting with human beings, even random human strangers, is more psychologically nourishing than connecting with a chatbot. That said, connecting with a chatbot is far easier. It's friction-free. Chatbots will support your every idea, tell you that you're a genius. If you have a conflict with someone else, you will always be on the right side. They're totally obsequious. They're obsequious, sycophantic. And more than that, they can't ghost you. So I think that the same way that hyper-processed calories are easy to come by and sometimes taste good but are not nourishing, so too AI and synthetic relationships are abundant.

12:55Jamil Zaki: They give us that sort of dopamine hit of being approved of, of not taking risk. But ultimately, do they nourish us? I think that's an open question for the long term. In the short term, it appears that they don't.

13:07Bryan Hancock: Do you think they inculcate us at all in how to treat other people empathically because they are obsequious and sycopantic and so forth? Do you think there's any learning in that if people experience them as empathic?

13:24Jamil Zaki: Yes, I think that. So let me first, because I realize that I'm sounding like a little bit of a doomer. Let me be clear. I think that AI is an astonishing technology with incredible upside. And I think that upside includes helping people connect with each other more effectively. With that preamble out of the way, yes, I think people are learning. And I think they're learning the exact wrong things from AI right now. There was a fascinating study out of Stanford, in fact, right here, that found that when people talked about a conflict that they had been in with a sycophantic AI bot, they were less likely to say that they wanted to make up with that other person, to apologize, or to reconnect with them.

14:09Jamil Zaki: If they instead interacted with a bot that challenged them, they were more likely to say that they wanted to make up with the person, but less likely to say they wanted to interact with that bot again. People like to be approved of. They like to be praised. But what they learn from those bots is, I was right all along, and I don't need to make amends. I don't need to reconnect with that person. That, to me, is the exact wrong lesson. Is there a set of settings that we should all enter into our preferences and claw to whatever to say, you know, don't be sympathetic. Don't suck up to me. Don't make it all that I'm right.

14:47I want the no BS filter. Have you seen that? Be successful.

14:52Jamil Zaki: Do you do it yourself? This is exactly what researchers in these studies do. They set bots to either be sycophantic, neutral, or challenging. And the challenging bots are the ones that I think have two positive side effects. One, they make us more open to other people's opinions and perspectives. And two, they make us less inclined to stay on platform, right? Because if a bot is challenging, you're like, okay, I get it. Now I'm going to go. I don't really feel like talking to you anymore because you're stressing me out. But I do see the error of my ways potentially, right? And I think that that's really useful in the exact opposite of what most frontier labs are trying to do.

15:31Jamil Zaki: Frontier labs don't optimize for people learning from their products or reconnecting with other people. They optimize for engagement. In other words, what will keep people on the platform for as long as possible? My own cloud settings are, you know, they're not be mean to me. Please show me the other side of the argument that I'm making, you know. And it does so very effectively. And I think all of us can do that. Right now, though, I think we're still in many companies talking about agents as if more agents equals more impact. And I think that is simply not true. And actually, fewer agents that are deployed in more strategic ways, I have found is actually, especially if it's part of a larger reimagined workflow, is a much more stable and viable path to impact and value.

16:22Jamil Zaki: Yeah, I think that's right. And, you know, there is this kind of kitchen sink approach that a lot of companies are taking to AI deployment. And why not? I mean, I get it. There's this once in a century technology. It seems like everybody else is using it. You have FOMO as a leader and you say, we must AI. Right, right. And your board is asking you too, what are you doing on AI? shareholders, board members are all exerting this enormous pressure on you to show that you're doing something. But I think that sometimes speed is the enemy of effectiveness. And I think we're seeing that in AI deployment really across lots of different companies where AI is being deployed in a haphazard way, as you said, where the assumption is that more equals better.

17:06Jamil Zaki: And that also, to your earlier point about social capital, I think it means that a lot of leaders are trying to deploy AI in a messy way while also not really communicating clearly with their employees about what that means for them. And if in an AI deployment, if in haste to deploy as much as you can, you lose the trust of your employees, I mean, I can't guarantee, but I would suspect strongly that this will be a net negative for your productivity, not a net positive. Yes. I mean, one of the things I think is interesting is how do you start to control the communications, what it looks like as a leader that's rolling out, that knows that trust is important.

17:48I've seen one organization feed every speech that a leader has made through AI to determine who's off message. And so it's using AI to help reinforce trust, not in the way that you might think, but in the way of monitoring the communications of the other leaders and then going back and providing feedback saying, hey, if you're talking about it this way, that may not be optimal for what we're going for, you know, reframe it.

18:16Jamil Zaki: That's fascinating. That's a really interesting use case. And it's one where AI seems to be sort of more in the role of a coach than a crutch, I guess, you know, I mean, I think that if we're using AI to just replace what I mean, you could imagine, okay, now we're going to have, you know, chat GPT or Claude write your speech for you, that would maybe be more of a crutch. And I think that if employees were to know, hey, my leader seems to be trying to inspire me, but I know they're just reading off of a bot transcript, that I think would decrease trust. But if you're using it instead at a top level to say, hey, here's one thing that's missing.

18:52Jamil Zaki: I leave it to you to figure out how to get back on message, but I'm giving you an alarm signal. That's sort of more in this challenging framework as well as opposed to a sycophantic one. So that's actually, it strikes me as a positive use case socially so long as AI is not then used to replace or script exactly what a leader says, but rather to just alert them for where they should be pointing their human energy. How do you build trust in an environment where some people truly will need to reconcile with their role, either fundamentally changing or perhaps going away? Oh, gosh, so much to say about this.

19:30Jamil Zaki: I mean, look, technology and technological shocks will always change the workforce. What I've seen a lot of is silence from leaders, right, where they're rolling out AI and the implication is you are going to be replaced, but no one even says that. And that, I think, is a great way to destroy trust, not only among the people who might have to lose their jobs, but among the people who will stay on. right because if you just laid off a bunch of people and didn't tell them anything about it and you didn't lay me off well i could still be next right so i might be looking for the exits i think it's critical to communicate as clearly and as early as possible about potential impacts when it comes to voice i think that that having people at least be able to speak up about their experience and the way that they think they can interact with ai can be useful and in terms of support, I mean, I think that, yes, if layoffs have to happen because of AI efficiency, we can think about, well, how do we support people?

20:28There's a bunch of evidence that layoffs, when

20:31Jamil Zaki: handled in this way with procedural justice, generate less bitterness from people who are laid off and less of a hit in morale for survivors who are not laid off. So who within an organization would you see best holds the keys to procedural justice? Ultimately, it's a CEO and the entire C-suite that is deciding what direction to take. I think that one question that will always come up if layoffs occur is, did the leaders of this organization look for alternatives? Did they think about retraining people who were potentially on the, you know, who might be laid off instead? Did they think about ways to use augmentation across the firm?

21:13Jamil Zaki: If they did, then those layoffs will be perceived very differently than if it appears that it was the first option. If leadership was just looking for a way to downsize and then the CHRO is put in the position of trying to reframe all of that through the lens of procedural justice.

21:29Bryan Hancock: Jameel, you spoke earlier, you invoked CEOs and their shareholders. And I think all of us would agree that empathy and human connection is a plus. But CEOs are presumably not encountering empathetic shareholders on earnings calls and so forth. So how seriously do you think CEOs do or do you see them taking this as an investment with so many other dynamics in the operating environment right now?

22:02Jamil Zaki: I think there's what I'm seeing and what I wish I was seeing. So Business Solver has this great yearly survey, the state of workplace empathy, where they survey CEOs and their employees about all things empathy related. They've been doing it for 10 years. CEOs in particular said that they didn't think empathy was as important as they did just one or two years before. They also reported having less time to connect with their employees as human beings than they did just one or two years before. So this technological advancement seems to be coinciding. I'm not saying one caused the other because we can't determine that, but it's coinciding with a decreased investment in human beings, according to this major survey among CEOs.

22:45Jamil Zaki: So maybe they're feeling this shareholder pressure and saying, hey, it's time for AI. It's not time for people. That, though, is, in my estimation, a major mistake, both at the human level, but also at the level of AI rollouts and business itself. As we see in lots and lots of surveys and studies now, when employees don't feel as though their leaders care for them, they are less willing to innovate, less willing to use new technologies. There's a massive gap between what CEOs think their employees are doing with AI and what's actually happening. I think 40 % of CEOs in this big survey said that they believed that the average employee at their firm had saved at least eight hours a week through the use of AI.

23:29Jamil Zaki: But two-thirds of their own employees said they had saved two hours a week or less. Also, if you pressure people to use AI without sort of empathizing with them, without building a relationship with them, they use it in these very suboptimal ways, producing work slot, low quality, low effort outputs that actually take people more time to work with. And in another survey, 44 % of Gen Z employees, these younger folks that we were talking about, admitted to purposefully sabotaging the AI rollouts of their organization. So on the one hand, I think CEOs in many cases might be caving to this pressure of speed above all.

24:07Jamil Zaki: And I think that might actually in the medium to long term hurt them in terms of the speed and efficiency of their AI rollouts. I completely agree. And I think the tide is already shifting on this one. And I think a couple of things are coming together. One, there is more focus on the value of AI. So we see much more talk about I'm investing this much in AI. What is the return on that investment? And so I think that focus on value raises or puts a spotlight, let's say, on uses of AI that are more in the slop category versus truly value adding. I also think more research, including your own, is coming to bear that shows that trust is foundational.

24:52And all this, McKinsey also found in a recent report around organizational readiness for AI, that trust was a foundational element. And in some ways, it's like it sounds obvious. But yet, to your point, it wasn't actually a centerpiece of many AI strategies in the past couple of years. So I don't think I'm being Pollyannish when I say I observe that things are starting to shift in the direction that you want more organizations to go in. And I agree with that kind of as a goal because of some of the factors that we're seeing shift now, the economics and then also just the feedback on the approach.

25:32Jamil Zaki: Well, first of all, that's very encouraging to hear, Brooke. I'm very happy to hear that. It's also astonishing how quickly this is all changing. Now it feels like we're in the third generation of AI adoption. It's been just that many years. But that's great. And that shows real learning. I mean, the data are very clear at this point that when people feel psychologically safe, they innovate more. Same goes for empathy. If you feel that, hey, my leaders have my back, well, then you get to take more risks. You get to be more creative. And I think you see this when you look at companies that are producing real AI maturity.

26:10Jamil Zaki: It's because they're asking their employees to be their best selves using this technology as opposed to threatening them silently or otherwise with replacement.

26:21Bryan Hancock: Jamil, you mentioned that great anecdote about Gen Z sabotaging their organization's rollout efforts. And it made me think, is there a generational component to empathic expectations, at least? Do Gen Z employees expect something different from other generations? And how is that to be navigated, if so?

26:47Jamil Zaki: Absolutely. The business solver has found year after year that Gen Z employees feel that empathy from their leaders is more important than, you know, Gen X, for instance, my generation or older generations. They also are much more likely to be disappointed in the amount of connection that they feel in the workplace. I mean, it's also true that Gen Z is lonelier as an overall generation, not just in the workforce, than older generations. So you see this kind of social hunger and social disappointment among that group. And I think it's, I mean, I feel a lot of concern for younger people in the workforce now.

27:25Jamil Zaki: I mean, it just must be such a difficult time to be trying to map out, chart your future and your career with so much uncertainty and so much disconnection. And I think that it's really up to leaders, up to our generation to try and figure out the best way to support younger people. There is something about AI, too, in which we are all being disrupted.

27:50Bryan Hancock: We are all having to rethink what we do, how we do it. And that changes the traditional power hierarchy to some degree, right? Does it help at all that we are going through this with young people? I mean, does that conduce to empathic learning that we're all having to figure it out at the same time rather than the traditional, here's how it's done, do it this way or don't do it at all?

28:18Jamil Zaki: I think it depends on whether we actually are going through the same thing and whether we're clear about the ways that we are and we aren't. You know, I recall during the early pandemic, leaders would say, hey, we're all in the same storm together. And their employees would say, yeah, we might be in the same storm, but you're in a giant yacht and I'm in a little 50-year-old 12-foot sailboat, you know. So in that, yes, we were going through this disruption. But if you're a leader, you have job security. You might have the resources to hire extra child care. But you might live close to where, you know, to all the different places that you need to be.

29:00Jamil Zaki: Whereas if you're an early stage employee, you might not have any of those. Your vulnerabilities are much greater. So, yes, I think the spirit of what you're saying, I accept, I think, is really powerful. We are all going through this change together. But I think it's important not to overuse the language of similarity when, in fact, we might be experiencing this very differently. So how are like four year institutions, you know, whether they're the big iconic Stanford's or not helping to create the conditions for that? It's a great question. And I think that there's been extreme divergence in the way that educators have responded to this revolution, right?

29:41Jamil Zaki: A lot of my colleagues, especially those in computer science and engineering, are really helping their students use AI in all of their processes to be much more efficient. I think that this technology has really forced me and many of my friends to question, what is college for? you know because many of the skills and the knowledge that we if you think of college as just transmitting a bunch of information from your head into somebody else's you should close all colleges now because that information is now available to them in whatever moment they need in whatever format they want right uh i see college as and i always have as a place to learn who you are at a deeper level and to practice certain types of deep thinking.

30:31Jamil Zaki: I personally am biased as an author. I think writing is an extraordinarily powerful way to think and to become a better thinker. And so in all of my classes, I've, in essence, outlawed the use of technology for writing. All of my lecture halls are tech-free. Students take notes by hand. In my seminars, I give all of my students moleskin notebooks and I do writing exercises. I've moved away from assigning writing exercises by computer, where I think the temptation to use AI is just so great for a busy student. And instead, I say in my classes, large classes, we do blue book exams. Do you remember blue books?

31:10Jamil Zaki: Oh, yeah. Yeah. And, you know, I have to deal with a lot of poor penmanship and try to decide. I go cross-eyed reading what my students have written, but I prefer knowing that they wrote it. and frankly watching them write it. And I think that there's a lot of pushback to that, like why, what is the point of a blue book exam? And to me, it's not to give everyone carpal tunnel syndrome. It's to help students understand that writing is a creative thought process. It is not merely a translation or communication process. And what you're describing lines up very nicely with what some of the CHROs I'm talking with now that are actually increasing their hiring of entry level folks is what they're saying they're hiring for is less the deep technical knowledge that they learned in school.

31:58They are less concerned that that transmittal of knowledge happened. What they're more concerned is that somebody is curious, that they know how to think well, that they know how to communicate and empathize with peers. And so they're looking in some ways to have a more complete general athlete, somebody that learned a bunch of things in college outside of the specifics of whatever underlying field they were studying and more how they became a more full human that then can use AI to supplement some of the technical bits that they may have or may not have fully picked up because they're interested in the human parts.

32:38And it sounds like the way you're doing it in class very much lines up with some of what those employers are looking for.

32:43Jamil Zaki: Well, I'm glad it does. And I'm really glad to hear employees thinking in these terms because, I mean, to me, that is the aspirational model of any technology is that it allows us, forces us really to ask the question, what are people good for? And then to point our efforts towards those unique qualities. You know, we no longer need to be information aggregators, right? But what are we for? So if we can take this moment to say, yeah, I'm going to offload the processes that, you know, that I don't want to, that were never my strength to begin with, like doing a bunch of calculations or storing as much information about virology in my head as possible, and rather focus on how can I be, how can I do work where I am applying my creativity, my taste, my ability to learn and my ability to collaborate with others.

33:38Jamil Zaki: that's an ideal model of a technologically enabled workplace and life in my opinion

33:44Bryan Hancock: jimmy alzaki thanks so much for joining us this has been such a pleasure and welcome to stanford

33:50Jamil Zaki: everyone i hope that you enjoy the rest of your visit here great thank you so much thank you great to see you as always absolutely it's a fun conversation very fun