Sergey Kravchenko - Fotolia

Ret. Gen. Stan McChrystal on what's wrong with modern leadership

Retired four-star Gen. Stan McChrystal talks about how modern leadership needs to change and what leadership means in the age of AI.

Retired four-star General Stanley McChrystal, who led the U.S. campaign in Afghanistan from 2008 to 2009, has spent the last couple of years on a different campaign: rethinking leadership in the modern age. After years of military command experience, he realized that modern leadership training falls short by focusing on the individual at the top.

McChrystal, 64, left military service after a Rolling Stone profile created some friction with the Obama administration. This led to a re-evaluation by McChrystal of why leaders succeed and a new book, Leaders: Myth and Reality, written with co-authors Jeff Eggers and Jay Mangone.

SearchCIO contributor George Lawton met McChrystal at a launch party for his new book in San Francisco and later caught up with the general by phone. The conversation here focuses on McChrystal's views on modern leadership, including his perspective on what leadership means when machines are making life-and-death decisions.

McChrystal is a senior fellow at the Yale Jackson Institute of Global Affairs, where he teaches a course on leadership. He is the founder of McChrystal Group, a consulting firm focused on leadership development and improving organizational performance.

Editor's note: This interview was edited and condensed.

The Rolling Stone profile

What did you learn in the aftermath of the Rolling Stone story about modern leadership?

Gen. Stanley McChrystalGen. Stanley McChrystal

Gen. Stanley McChrystal: The Rolling Stone story happened in the spring of 2010. I had a series of reporters doing stories on the war in Afghanistan. And this reporter pitched the idea of doing a story on the command team, and we thought it was going to be kind of a puff piece. He was only around for a few days on and off over the course of six weeks. When the story came out, it basically had a theme that my command group was a little bit like a locker room ... not very respectful of Vice President Biden and irreverent and that sort of thing.

It hit at a moment when there was this perception that there was a military and civilian divide in the U.S. government. This really struck a chord with a lot of people so that the story kind of blew up. I didn't think the story was accurate, but that was irrelevant at the time. When the story came out, I went to the president and offered my resignation. I said, 'Here's my resignation. I'll stay if you want me to, or I'm happy to resign,' and he was very gracious. It was a good conversation. He accepted it, and my military career ended in that moment. The big lesson learned was not the obvious one to just stay away from the press or that sort of thing.

The big lesson for me was: What do you do when something like this happens? Because, for me, this was entirely unexpected. I had thought I might be killed in combat, fired for incompetence or any number of things. But I had never, ever in my wildest dreams, foreseen a scenario when I would be accused of being disloyal or undisciplined. If I had spent the rest of my life feeling sorry for myself, then all I was going to do was waste the rest of my life. The point is: Everybody's going to fail in some way; that's largely unavoidable. But what is not unavoidable is how you react to failure and how you interact with the people around you who fail.

In my case, I was suddenly out of the army. My staff weren't dependent upon me anymore. But they were dependent on me to validate what they were doing; most of them had gone to Afghanistan because I asked them to. Most of them have been extraordinarily loyal and hardworking.

I'd spent years with some of them, and I felt I had to demonstrate that I wasn't a bad person that was suddenly unmasked. What I had to do is try to act as much like a leader as I could in the future so that, when they looked at me, anything they read in the context of that story was just one data point.

'Followers,' situation determine leadership style

How did your thoughts on leadership evolve at this point?

McChrystal: I'd been taught leadership at West Point and then exercised it throughout a military career. Then, I wrote two books, my memoir and Team of Teams, exploring leadership. I came to the conclusion, along with my co-authors of Leaders, that we didn't really understand leadership because we had these ideas that, if you do certain things, you're going to be successful. People were saying that, if you have certain behaviors and traits, if you take certain actions and if you're bold, you will be successful.

Leadership is the result of the interaction between leaders and followers -- almost like a chemical reaction in the context of the moment. When you think about leadership that way, you realize that every situation is different and requires different leadership traits.
Stan McChrystal

But when you look at the data and you look at history, often, people who do everything right lose, and people who do nothing right succeed. So, you look at that, and you say, 'Wow, there's something else to this.' Then, we thought that maybe leadership is not this thing that someone develops as a skill or is born with.

Leadership is the result of the interaction between leaders and followers -- almost like a chemical reaction in the context of the moment. When you think about leadership that way, you realize that every situation is different and requires different leadership traits. Just because you were effective in one place doesn't guarantee you will be effective again. Although, if you're adaptable, you might be good at reading the situation and doing it well. The biggest revelation was that followers are a bigger part of the process than we really thought, because followers are individuals that each respond differently. They are a group, and when they are networked, they have a group response that is like behavioral economics.

Everybody can appreciate behavioral economics, but you can't treat it like a finite science because you can never predict how people are going to respond completely. Leadership is much more like that because you are dealing with this group of followers -- although I don't even like the word follower anymore. I think participants or partners are better terms. What you're doing is you're interacting with this thing and it's constantly changing. You can think of it as surfing. There are no two waves that are the same. When we got to that point, we started to say, 'This is really significant because how we select leaders, how we promote leaders and how we try to lead are all impacted by our understanding of what are we really dealing with.'

What can someone do to be a better follower in light of this idea? For example, let's say you see a leader is making a mistake, and you feel you need to say something but don't want to undermine the leader's power -- or tarnish the symbol of power. What are some of the ways of cultivating that balance between respecting the leader and speaking up in a way that grows the organization?

McChrystal: Followers have a huge responsibility or agency for what happens. Followers can do several things. One of the first things they can do is just demonstrate the behavior that they think they want the senior leader to have. And it doesn't mean they act like a senior leader, but those leadership things that they think are appropriate with this group and situation. When they demonstrate them, it's really a subtle way of showing the leader, 'Here's what we need right now, boss.' Whether it's compassion or being a hard-ass, or whatever it is.

Also, followers can set standards in an organization. Sometimes, the leader comes into an organization and says, 'This is what I want.' The followers can say, 'Well, boss, this is what we do here, and this is why.' They basically are informing this leader about what are the norms and what is the culture there. In many cases, the culture is very strong. It's a good thing. So, the leader gets exposed to that and thinks, 'Wow! I'm not painting on a blank canvas here. I'm painting on a canvas that I need to be respectful of and get the best out of and change only where it's appropriate.'

The last one is probably the hardest. That's talking to the leader and telling them to change. It's easy to talk about telling a leader they need to change some things. But, when you're more junior, it is unrealistic to expect most people to do that comfortably. And many senior leaders wouldn't respond well to it if they did. So, what I would say is: Senior people have a big responsibility to do that. But I think you're better to do it more subtly.

Empathy and modern leadership

What skill sets can leaders cultivate to become that touchstone or symbol that inspires followers?

McChrystal: The leader's got to be able to communicate so people can understand and be inspired. Really effective skills for a leader involve empathy. A leader senses what is happening in the environment and what is happening in the followers; the leader understands the mood and the need. I know from the military that you can tell what is going on, even over a radio, where you can't see someone. You can tell in a firefight how a junior officer is feeling because it comes through in their voice. [Expanding on that], the leader with a willingness to spend time trying to sense others knows when people need a pat on the back or to be hit on the head. They must be willing to adapt their technique to what's appropriate for the moment.

There are moments where the leader is far more effective doing one thing than the other. Even a tough guy leader sometimes has to hug people. Sometimes, a very compassionate, jolly leader needs to rip into people. So, it's that ability to read situations with humility then adapt yourself and to not use a hammer when a screwdriver is the appropriate tool.

Modern leadership in the age of AI

What can leaders to do adapt to the rise of AI in order to effectively hand over more decision-making power to machines?

McChrystal: This is something we've done a lot of thinking about. I don't claim we've got any perfect answers, but here's what I'm thinking about it right now. If you think about what machines do well, they can take a lot of data and, given certain parameters, can then make sense of that data and make decisions. They have to be programmed to make decisions to respond to certain things in certain ways. From the standpoint of management, you could argue that, with AI, the machines could do a heck of a lot of management better than people because you can connect them to more information input. You can give them very complex algorithms and do things like speed up the assembly line.

In the military, people in a meeting would say that we should never have machines involved in a lethal loop without a human being involved. I know why they're saying that, but they're completely wrong when you think of the requirements for war. For example, we have missile systems that protect naval ships, and the big threat to naval ships is high-speed missiles that are shot at them that come in at hypervelocity. If you try to put a human in the loop to fire these defensive systems, they won't be able to get the data and make a decision fast enough, which will cause failure. So, you've got to automate that.

Now, the challenge is, when you automate that, you've got to be so good at automating that that the machine approaches human judgment, which can also make mistakes. Back in 1988, the U.S. ships in the Persian Gulf saw an Iranian aircraft take off, and it was approaching in the direction of the ship. They engaged the airplane and shot down a civilian Iranian airline with almost 300 people killed. There was a person in the loop responding to the information that was being provided by sensors.

You're going to have to get your AI so good that it's like the autonomous driving vehicle. We need to set up the ethics that the machines are going to be governed by. I don't think we're close to doing that, but it's going be critical to set up the ethics of the algorithms. And we are going to have to understand the problem like never before. We are going to have to understand it from inside out, because you're going to have to try to tell something as dumb as a machine how to think about it.

What's a good framework for thinking about cases where machines, like self-driving cars, might kill people by accident but at a much lower rate than human drivers?

McChrystal: Max Weber came up with the concepts of bureaucracy a century ago, and it was designed to have decision-makers execute things according to rules that don't let emotion or corruption get in and pollute the system. And we all hate bureaucrats, because you go to a bureaucrat and say, 'I want to get a driver's license, but I don't have the paperwork,' and they say, 'Sorry.' In the moment, you're enraged and hate them. The machines are going to be like that. They are going to be like bureaucrats with no judgment unless we program in the judgment.

How are we going to teach machines to get the higher-order thinking and use what we would think of as judgment to do [their jobs]? There is an endless number of things that they cannot handle, but humans can. And that is going to be a key. We already feel the pain of automation. For example, when you go to a website and you can't do something. You would like to talk to a person, but that's going to be less common. This means the reaction of the AI in the machines behind the apps and websites is going to have to get a higher level of flexibility and judgment.

Dig Deeper on CIO strategy

Cloud Computing
Mobile Computing
Data Center
and ESG