The McKinsey Podcast

Gen AI talent: Your next flight risk

| Podcast

Employees who regularly use gen AI are likelier to be more productive, more efficient—and more attracted to your competitors. On this episode of The McKinsey Podcast, McKinsey talent leaders Aaron De Smet and Brooke Weddle talk to global editorial director Lucia Rahilly about what these workers want most, as well as practical steps leaders can take now to keep them happy and engaged.

In our second segment, McKinsey partner Julia McClatchy discusses why CEOs must be their company’s chief storyteller. This is the start of a new series we’re calling CEO Insights, featuring short, sharp perspectives about the evolving role of CEO.

This transcript has been edited for clarity and length.

The McKinsey Podcast is hosted by Roberta Fusaro and Lucia Rahilly.

AI talent is primed to quit

Lucia Rahilly: Gen AI is obviously a juggernaut with the potential, arguably, to reshape work as we know it. It’s early days, but gen AI skills are already in demand in many fields. Yet your new research shows that talent skilled in gen AI is more likely to quit than the rest of us Luddites. Brooke, tell us about this top-line finding, and why employers should care.

Brooke Weddle: People who are most fluent in some of the skills required to engage with gen AI, to build use cases around gen AI, are flight risks at this point. And that’s an important finding, given that many companies are trying, relative to gen AI and a broader set of technological capabilities, to upskill and reskill in-house talent versus acquire new talent.

Lucia Rahilly: And just for context, what constitutes “gen AI talent”?

Aaron De Smet: It’s important to understand that there are different types of gen AI workers. We categorized four types. In one group, 12 percent of respondents said they would consider themselves technical employees. They are creating gen AI and next-gen artificial intelligence platforms, programs, and software. But most people who use gen AI don’t even consider themselves technical workers. In the biggest group, 88 percent of respondents said they don’t consider themselves technical employees.

Lucia Rahilly: So most folks who consider themselves gen AI talent use gen AI to gain efficiencies.

Aaron De Smet: That’s right.

Lucia Rahilly: What would this kind of nontechnical use look like?

Aaron De Smet: For example, a corporate-communications worker in a large company might craft a lot of press releases or emails from top executives. They often interview a CEO who’s about to be featured in a town hall, or about to send out an important message. They’ll need to synthesize the leader’s thoughts into an initial speech. That work of a communications professional could take hours. But with gen AI, they can focus on other things like messaging and being more strategic in their thinking.

Brooke Weddle: The McKinsey Global Institute did research on the most commonly deployed use cases in gen AI right now. And the most regularly reported gen AI use cases by function are in marketing and sales. So gen AI is creating personalized marketing and drafting technical documents, and in service ops, the chatbot function is able to answer customer queries. Most people think that gen AI is used in highly technical kinds of instances, but if you look at the frequency of these use cases, we’re seeing that it’s much more attached to, in some cases, mundane tasks.

Want to subscribe to The McKinsey Podcast?

The risks of not listening

Lucia Rahilly: More than half of these gen AI workers—those who self-identify as creators and heavy users—present a flight risk to employers. Leaders want to keep these folks not just employed but also productive and engaged. Tell us about the stakes of disengagement to companies.

Aaron De Smet: The stakes are huge in two senses. One is that employees have tremendous potential to be more productive with the application of gen AI. When disengaged, that productivity goes down markedly; it is often cut in half. Their productivity is a mix of working less and working fewer hours, but also the quality of their work goes down as their engagement goes down.

The second thing is, they just leave the company. And if they were good at their job and really good at using gen AI, you’ve just lost very valuable talent. And the company will not only have to replace that talent, but if the reasons why they left are systemic in the organization, there might be more attrition coming with the new talent that is hired.

Lucia Rahilly: Makes sense. Where should leaders double down? What did the research tell us about what gen AI talent cares about day-to-day?

Brooke Weddle: The research was very clear on this front. Compensation is not the driving factor here. It is about having flexibility, meaningful work, reliable and supportive colleagues and teammates, and a clear sense that the employer is focused on health and well-being.

I think it’s a mistake to think that companies can move away from having this emphasis on meaningful work and well-being. Leading companies I’m working with say, “Yes, we need to get to more productivity and productive outcomes. But we’re going to do that by understanding holistically what employees want and knowing it is not just about financial incentives.”

Compensation is not the driving factor here. It is about having flexibility, meaningful work, reliable and supportive colleagues and teammates, and a clear sense that the employer is focused on health and well-being.

Brooke Weddle

Lucia Rahilly: How is it possible that compensation has fallen as a factor?

Aaron De Smet: It’s extremely important, but just more binary. A lot of people wouldn’t even look at a job to begin with if it didn’t meet their minimum compensation requirements. But once it meets their minimum compensation requirements, paying more doesn’t get you a lot. In some ways it’s probably still the most important factor, but it’s binary in a way that many of the other factors are not. As we mentioned, the more talent feels part of a supportive community of collaborative colleagues, the better.

Lucia Rahilly: We’ve heard a lot, as you alluded to, Brooke, about flexibility in previous iterations of this research, particularly during the pandemic. Just coming out of it, what’s different now about the way flexibility might matter?

Brooke Weddle: It’s about having a greater sense of control and not just about, “Am I in the office or not?” I think that is a simplification that hasn’t been entirely helpful. It’s about reimagining shifts at work, reimagining whether or not you need to do work with someone else versus on your own.

I’ve been intrigued by some of the changes that large traditional manufacturers are making when it comes to understanding how to meet their needs and get older talent into the workforce, and from a childcare perspective, how to get moms into the workforce in ways that meet their needs.

There’s been a real broadening of the way in which we think about flexibility that should apply to, and can apply to, tech and nontech workers.

Aaron De Smet: It’s not just hybrid or remote work. It’s also work schedules. Do I have some autonomy over my schedule? We saw during the pandemic that a lot of people went from a double peak day to a triple peak day. People want to be able to set some boundaries and have some control and autonomy over their schedules.

As gen AI takes over routine tasks, this is one way to think about it: the work left for humans to do is more human. People want to be treated as humans. So in addition to flexibility and some level of autonomy and well-being, one of the other biggest disconnects between employers and gen AI talent—and, frankly, talent broadly—in terms of what matters, is feeling valued by leaders and by the organization.

One of the other biggest disconnects between employers and gen AI talent—and, frankly, talent broadly—in terms of what matters, is feeling valued by leaders and by the organization.

Aaron De Smet

Lucia Rahilly: How amenable do you find your clients to enabling this kind of autonomy and flexibility?

Brooke Weddle: It varies. The companies that have done a good job at figuring it out have a very clear connection with the business they’re trying to run and the work environment that they are creating to enable that. I think companies get into hard spots when they say, “You’ve got to be back at work, in the office.” And it’s not clear to everyday employees what that adds in terms of driving the performance outcomes of the company.

The majority of employers continue to think about it and continue to try to understand how to make those work environments more conducive to what employees need. They’re balancing that with emerging concerns around culture atrophying, especially with the younger talent who have recently entered the workplace. Are they, for example, getting the apprenticeship they need? Some changes regarding bringing people back into the office have been well intentioned in trying to address that, while also recognizing that their employees need independence and autonomy in the way in which they work.

Recruiting gen AI talent

Lucia Rahilly: So suppose I’m an employer, and I’m looking to pick up some of these gen AI workers who are poised to leave their current positions. What attracts folks most to a new employer?

Aaron De Smet: Some of the factors that make the biggest difference are having caring leaders, meaningful work, flexible work, feeling a sense of inclusivity and community. These are some of the biggest factors that are making a difference in terms of attracting and retaining this kind of talent.

Lucia Rahilly: What do you make of the fact, when it comes to attracting talent to a new job, that flexibility dropped from the number-one slot in what employees cite as a reason to stay to around five or six on this ranking.

Aaron De Smet: People are looking for a place where they can feel appreciated, supported, inspired, connected. If they don’t have that, then the flexibility doesn’t matter as much. I think flexibility is becoming table stakes.

Redefining work

Lucia Rahilly: There’s a lot of general Tom Sawyering that needs to happen to get teams of human employees to produce great work. Whereas if that work is shifting from, for example, junior-level employees to AI, not so much, right?

Brooke Weddle: For so long, we’ve had a pretty steady definition of what I’ll call “hard work,” which is defined as paying attention, taking notes, summarizing, and sending [the notes] around to coworkers and/or executives. This enabled managers to tell if someone was doing what they needed to do. It showed managers that employees were putting the time in, whether it was research, summarizing, synthesizing, and doing something productive with it.

Now with gen AI, the question becomes: How do you define hard work? If a colleague is able to use gen AI to rapidly get to what would otherwise have taken perhaps hours, is that still considered hard work? As we think about performance management, we should first give hard work a new definition; second, think of a different way of measuring and managing that; and third, think about what the aspiration is—is it more of the how versus the what?

Aaron De Smet: I agree. I think the big shift is: How do we measure productivity? I used to look at output. But now output for a lot of these routine tasks can be done in seconds by AI. I can’t measure it by hours worked, or lines of code written, or a lot of the other traditional measures of productivity, because those don’t apply in the same way anymore. Now I have to look at the quality of the work and the impact of the work.

Let’s take the work of a songwriter. What if they wrote one amazing song in an hour? Would you say, “You only wrote one song, and it only took an hour?” But if it becomes a hit you would say, “That’s much better than the person who spent hundreds of hours writing 100 songs that nobody likes.” We are not good at performance managing things that are that ambiguous. But the things that are easy to measure, easy to observe, easy to count, those are things AI is taking over. It’s just a whole different game. And I don’t think we know the answer to the question yet.

Lucia Rahilly: Any sense of how gen AI might alter the way we think about spans and layers in the org hierarchy?

Aaron De Smet: I think in a lot of cases it won’t change. For those who had very broad spans because the work was so routine, I think it will change, because that routine will now be done by AI, analytics, robots, and machine learning.

Immediate steps leadership can take

Lucia Rahilly: Leaders are confronting a scenario where the same employees who are most valuable to retain are also the likeliest to leave. If there are one or two things employers might do now to try to up the odds that their best workers will stay on, and stay productive, what would they be?

Aaron De Smet: First, create a healthy work environment that emphasizes both productivity and performance, sustainable work practices, and well-being. Second, invest in leaders at all levels. And think of every employee as a potential leader. That will help to create a strong, supportive community that can help people be at their best and more sustainable in the work they do. They can help one another in a collaborative way and be more productive, not just individually, but as a team.

Brooke Weddle: I’d add two things. One, in the research, I was a bit surprised by the number of nonusers. I think to the extent that it makes business sense, companies should upskill, reskill, and broaden access to gen AI–based tools and use cases. It doesn’t seem sustainable, given how fast we are moving on this front, for there to be such a small number of heavy users and creators.

The other thing I would do is think holistically about your change management plan. When people move to using gen AI, it saves a lot of time. How can you then help people understand what to do with that time they have created, and to use that time productively? I don’t think people are necessarily going to be able to intuitively create more productive outcomes without managers and leaders shaping that space with some intention. So that is the other push I would make, which is just thinking through how work gets done, and how to make sure that time saved turns into productive time.

Gen AI is not the enemy

Lucia Rahilly: When I was reading this research, I found it surprising that respondents reacted in such an overwhelmingly positive way to gen AI, even if more than two-thirds of them aren’t yet using it at work. Only 4 percent of respondents expressed worry about job displacement, which I found incredible.

Aaron De Smet: What respondents are finding is that gen AI is freeing them up to do work that they would rather be doing in the first place. And as they lean into that work, they’re saying, “Wow. I’m adding a lot more value. My job isn’t going anywhere.”

I think there will be some cases where there will be displacement. But a lot of those are not for users of AI. For example, let’s look at a call center that has replaced humans with AI because it has cracked the code on 90 percent of questions that callers ask. Those displaced humans that picked up the phone are not the users of AI. That kind of displacement, I think, is still a concern.

But we’re tapping into users using AI to help them do a job they’re already doing. And what those people are finding is, “This is helping me. And it’s not going to replace me. In fact, it’s allowing me to put my skills to their best and highest use, by clearing out some of the simpler tasks that frankly weren’t that value additive to begin with.”

Brooke Weddle: Larger questions need to be addressed. And this is where I go back to change management and thinking holistically about the human experience as it relates to using gen AI in the workplace. Work is already changing. We’ve got to learn as we go.


The CEO’s distinctive storytelling capability

Roberta Fusaro: Next up, managing producer Laurel Moglen speaks with McKinsey partner Julia McClatchy about CEOs’ unique ability to tell their company’s story.

Laurel Moglen: Julia, why is a good story so important?

Julia McClatchy: It’s mission critical. Having a great story creates a virtuous cycle that enables CEOs to build a track record of consistency. It allows CEOs to take action, accelerate the idea, and make change in the organization. Second, it facilitates a mutual exchange between the CEO and his or her stakeholders to come up with new ideas and improve the business. It also creates an opportunity to pressure-test ideas, collect feedback, and improve strategy based on insights. Third, it gives something for teams to rally behind in a strong, consistent message. Our research shows that high-performing organizations are almost three times more likely than others to say that their organizations express narratives well.

Laurel Moglen: I’m so curious about that data point. Do you think it has a cascading effect where people just understand what their purpose is?

Julia McClatchy: I think it does. It creates this flywheel because if you’re a high-performing organization, odds are people from the very top to the front line know exactly what their role is and how they fit into the broader purpose of the organization. They understand their narrative, and they think their company expresses it well because it’s fueling them to do what they do every single day to drive value for the company.

Laurel Moglen: How should CEOs think about shaping their organizations’ story?

Julia McClatchy: The ability to tell a great story is grounded in a singular narrative that encompasses the CEO’s vision and their strategy for the future. CEOs should think about telling their story in two lenses. The first lens is being holistic in nature because their story must transcend multiple stakeholder groups, whether they’re talking to their board, the media, or their employees.

Beyond a great narrative for CEOs, we know that adult learning principles tell us that things must be communicated in different mediums again and again. Repetition and consistency are critically important; they do not come naturally to most people, but great CEOs master them.

Laurel Moglen: What are the key factors to consider when CEOs tell their story?

Julia McClatchy: In managing stakeholder relationships, the first step for a CEO to consider is their audience. The way that the narrative is delivered [cadence, depth of candor] should be customized based on the audience. For example, imagine the way a CEO might communicate with the media. It would differ greatly from the radical transparency that he or she might use when communicating with his or her board.

Second, given the pace of change and what is expected of a CEO as they evolve in a role, that narrative has to be renewed across various stages of the CEO life cycle. While a new CEO might be most focused on setting a bold direction, a more tenured CEO might need to focus on combating complacency or articulating a vision for the company’s next S-curve. This enables CEOs to multiply the value of their story over time as they grow in their role.

The third point is that CEOs need to inject a strong sense of the “why” into their narratives. This means linking whatever it is that they’re saying back to the company’s purpose so that it feels grounded. Ideally, CEOs should include some of their own personal purpose in that narrative, too. They can share their personal hope and passion for the company. This enables the best CEOs to move beyond that first step of influence, to inhabit the mindsets of their audience by injecting a lot of personality into their stories, which ties back to having a firm grasp on their “why.”

Laurel Moglen: Leaders have to pay attention to a variety of competing stakeholder issues. What do you find the best CEOs do well as they think about this balancing act?

Julia McClatchy: We see a few different archetypes of how this materializes and how they then balance opposing views.

I’ll share three examples. The first is sharing versus staging. You see this mostly on investor calls where CEOs have the unique ability to share what stakeholders need to know now while laying what we would call “breadcrumbs” to keep them coming back for more in the future. These breadcrumbs might be an allusion to growth or an allusion to future focus on productivity.

The second is business versus society. Stakeholders, whether that’s customers or employees, care just as much about what you stand for as what you sell. CEOs have to confidently represent the company, its values, ideals, and purpose. This means speaking out on topics that at times can be divisive and complex.

The third is self versus collective. They’re positioned uniquely to make communications a team sport as they establish and develop their own platform. If we take the analogy “a rising tide lifts all boats,” what we see is that CEOs make communications a core capability, so the CEO doesn’t have to go at it entirely alone.

Explore a career with us