The Global Cable, Power & Security, Technology Building A.I. Armies with Erik Lin-Greenberg

April 17, 2020
By Perry World House | The Global Cable

This week's episode of The Global Cable features Erik Lin-Greenberg, a Postdoctoral Fellow at Perry World House researching how the military adopts new technologies.

On today's episode, Erik talks about how his own military service has shaped his life and informed his research; what it was like to set up war games with national security experts to test how they'd react to new technologies being deployed on the field of battle; and the biggest challenges for the U.S. military as it attempts to harness the power of A.I.

Listen on Apple Podcasts | Libsyn | Spotify | Stitcher

Music & Produced by Tre Hester.

Franklin Few

On every episode of The Global Cable, we ask our guests the 'Franklin Few' - an updated version of a questionnaire used by Penn founder Benjamin Franklin. Here are Erik Lin-Greenberg's answers.

Someone you'd like to meet: Edward Lansdale, United States Air Force officer and CIA operative who pioneered clandestine operations and psychological warfare.

A book, movie, or anything else you'd recommend to listenersThe Cult of the Irrelevant by Michael Desch, which looks at the fast-vanishing influence of social science on national security. 

Someone who's recently done something that deserves praise or attention: Key workers like grocery store employees, truck drivers, and food industry workers, who are keeping Philadelphia and the country going during the COVID-19 pandemic.

Something Penn and Penn students can do to be of service to the world: Keeping in touch with and checking in on family and friends during the pandemic. 

Listen now.


Erik Lin-Greenberg [00:00:09] As I mentioned, I think the more important lessons for me from my time in uniform are those that deal with war on a more personal and emotional level. So having deployed to the Middle East myself, and having deployed airmen that worked for me, I think provided some really interesting insight into these very individual level views on the human costs and emotional reactions associated with much broader strategic level decisions on the use of force.

John Gans [00:00:41] Welcome to the Global Cable, a podcast from Perry World House at the University of Pennsylvania, where we discuss the world's biggest issues with people who work on them. I'm John Gans, the Director of Communications and Research here at Perry World House. Like many of you, we're working and recording remotely amid the global coronavirus pandemic. So this and other episodes will sound a little different. We appreciate your patience.

John Gans [00:01:05] Our guest today is Erik Lin-Greenberg, a postdoctoral fellow at Perry World House this year. Prior to entering the academy, he served as an active duty officer in the United States Air Force, where he led intelligence operations and continues to serve in the Air Force Reserve. His work has appeared in a range of academic and policy outlets including The Washington Post, Foreign Policy, and the Texas National Security Review. Today, Erik talks to us about the challenges faced by the U.S. military in using an artificial intelligence, how he ran his war games with national security practitioners to explore new technologies, and whether his own service in the military has impacted his research. Erik, welcome to the Global Cable.

John Gans [00:01:49] So you have served in the military and you are still a reservist in the Air Force if that's correct?

Erik Lin-Greenberg [00:01:57] Yes.

John Gans [00:01:58] So how did that military service shape your life and view of the military?

Erik Lin-Greenberg [00:02:04] Sure. So I think that the chance to serve as an Air Force officer really gave me the opportunity to gain firsthand insight into how the defense establishment operates. But I think more importantly, it taught me a lot about myself as a person. So in each of my assignments, which have run from the very tactical level—so serving in a squadron that conducted ICE-SAR or intelligence surveillance reconnaissance operations—to more recent assignments as a reservist at the Pentagon, I've really gotten to see how the U.S. military functions. So, what factors go into planning and conducting operations. But as I mentioned, I think the more important lessons for me from my time in uniform are those that deal with war and a more personal and emotional level. So having deployed to the Middle East myself and having deployed airmen that work for me I think provided some really interesting insight into these very individual level views and the human costs and emotional reactions associated with much broader strategic level decisions on the use of force. For instance, when the president decides to deploy forces, it might mean that individual service members—so that soldier, sailor, airman, Marine, or Coast Guardsmen—might miss a key life event like the birth of a child, or that they'll come home from a deployment or from conducting remote war fighting operations like those associated with the U.S. drone program and suffer from PTSD. So I think learning these lessons and trying to respond to them is a 22 or 23 year old lieutenant who frankly didn't know all that much, and I was in charge of this team of about 125 airmen, was incredibly eye opening and beyond tumbling.

John Gans [00:04:01] What stoked your interest in serving, like what was the driving force that landed you in the Air Force to begin with and how did your ideas and misconceptions and everything else change as you spent time in it?

Erik Lin-Greenberg [00:04:18] Sure. So I think like many people in my generation, I was deeply affected by the events of September 11th. And I remember being a freshman in high school and slowly the news of the events unfolded. I was growing up in a town right outside of New York City and I remember walking home that day after we were dismissed early and just seeing these fighter jets flying through the skies at really, really low altitudes. And so there was a ton of uncertainty in the world. But I felt a sense of safety knowing that there were folks watching us and protecting us and I wanted to be a part of that. And, you know, things worked out and I got to go through ROTC in college and serve as an Air Force officer. So, yeah, that that's how I got into the military business.

John Gans [00:05:05] And that's amazing. What was your understanding of the military before you went in and how did it evolve once you were in?

Erik Lin-Greenberg [00:05:13] Sure. So I think before I went in, I watched a lot of James Bond and war movies growing up and I think that was kind of my view of the military. But serving as an officer, I realized that, yes, your job is kind of the management of the use of force, but a lot more of it is kind of a human resources type role: taking care of individuals who've decided to put their right hand up and join the military. So I think that gave me a much better understanding of who serves in the military and the sacrifices that they're willing to make and then the challenge that they face as part of their service.

John Gans [00:05:51] So in getting to know that human face, that human side of the military, how does that affect your research and what are you focusing on these days and how did that service inform it?

Erik Lin-Greenberg [00:06:03] So I think a lot of my research focuses on military decision making and especially how emerging technologies shape military decision making and the use of force. One of the things that I often think about in my research as I try to set up research questions is, would my boss's boss care about the questions that I'm asking? And I think that's something that's incredibly important for me as someone who thinks social scientists should tackle policy relevant questions. The other thing that I think my military experience has helped me with in terms of my research is trying to find ways to tackle research questions in a way that actually draws from my military experience. So, for instance, trying to use war games in a manner that answers political science questions. And I think that type of research ends up being useful both for political scientists from a theory testing perspective, but also for policymakers who can understand what a war game is and can therefore more intelligently understand the information I'm trying to get back to them.

John Gans [00:07:14] That's fascinating. So if I could ask—you know, I've spent some time, both in the military and in the Pentagon itself, working as a speechwriter—and so my question is, how do you think that the military adapts to new things and new ways of thinking, new ways of doing things? Specifically technology. It seems to be you're researching it, the crux between decision making and technology. How does the military do when it comes to these things? And what was your experience when you were in uniform and what's your experience when you're looking at it from the academy?

Erik Lin-Greenberg [00:07:55] Sure. I think in most cases, the military does a pretty good job of developing processes and institutions and systems once technologies are already in the military. And that's why the U.S. Air Force, for instance, can still fly the B-52, this aircraft that first flew in the 1950s, because we've developed ways of upgrading it and ensuring systems are maintained and crews are trained. But I think what I've seen over the course of my time in the military and also studying it from the outside is where the military does a less good job is integrating new technologies. And I think there's a few reasons for this: the first is that we have this very antiquated and outdated acquisition system where something can take years to get from concept to actually deploying it out on the battlefield. And this is a model that obviously doesn't work for emerging technologies, technologies like software and A.I. that can leap year generations in short periods of time. So I think that's one issue. Another issue that I think the military is facing is that the entire military industrial complex is really shifting. So if we think back to the Cold War, the military industrial complex was made up of companies like Boeing, Northrop, Raytheon, where I think employees really expected to be working on defense related issues. But today, we now have companies like Google and Amazon that are potentially playing a part in the defense sector. And many of those employees don't necessarily sign up to be doing defense related work. And I think this poses some pretty vexing challenges. As we saw a few years ago with Google employees really resisting Google's involvement in a Pentagon sponsored A.I. development program. And so I think the military's trying to find ways to overcome these obstacles. They're really trying to streamline the acquisitions process. They've developed a number of new institutions, like the Defense Innovation Unit and Army's Future Command to again try to streamline the acquisition cycle. But I think more importantly, they're going through a pretty significant cultural shift in trying to make the military more appealing to the entrepreneurs and tech developers of today. So you have military officers now wearing civilian clothes instead of uniforms and engaging with the sector and folks really trying to take on kind of a venture capital and accelerator type approach to looking for new military innovations.

John Gans [00:10:32] So if I can ask a little bit, because I think that generally speaking, what we think about adapting to new technology is seeing new technology and acquiring it then eventually fielding it in military service. But I think one of the most challenging components of this is that you have a big gap between some of the big strategy questions and challenges that the country and the military faces. You know, how do you actually go to war? And what you actually have is, is that's usually done by your more senior officers. So 10 years in, maybe 15 years into their service, maybe 20. I think the chairman has been in 30 something years. But they tend to be, like anybody who's that well into their careers, probably far removed from technology.

John Gans [00:11:24] So how do you get the military leadership to adapt to technology and the opportunities that technology provides when they're actually developing strategies?

Erik Lin-Greenberg [00:11:35] Yes, I think there's a few steps to this. The first is to trust and delegate to younger leaders and to do a better job managing talent. I think one of the issues that the military faces is that, as you mentioned, it's this very hierarchical system in which the folks making decisions are much more senior than those that might often have the technical expertise. And I think we're seeing a small shift in military really relying on more junior individuals. So I have a friend, for instance, who the Air Force sent out to Stanford for a PhD—he's a captain in the Air Force—to really focus on development of new technologies and how to streamline these systems. And he's working essentially in the office of the vice chief of staff of the Air Force now to help develop these new systems. So I think that's the first step, is trusting these younger individuals and finding ways to integrate them into the military system. The second way is to start exercising these new technologies a bit more in war games and exercises. You typically think of war games and exercises as large force-on-force operations. But I think finding innovative ways to bring in these new technologies helps senior leaders realize both the benefits and limitations of these systems. So I think a combination of those cultural shifts and shifts in how we think about training and developing doctrine will be incredibly useful down the road.

John Gans [00:12:54] That's fascinating. So if I can push a little bit further, as you've been using your research and talking a little bit about using war games with national security practitioners to explore how they might react to new technologies that are deployed to the field like drones and things along those lines, are you doing that in terms of the adversary? Is that who you're thinking about? So, how do you think about a new technology once you see it beyond the wire? And then could you tell us more about how you set up those war games or what you learn from them? 

Erik Lin-Greenberg [00:13:28] Sure. So my war games, I typically recruit U.S. military practitioners to participate in war games where I expose them both to the availability of technology on their side, but also the adversary's use of technology. So do we see differential responses when, let's say, an adversary shoots that an aircraft, but that shoot down was either directed by a human or by an A.I. command and control system. And my research that I'm working on now, I draw from war games to figure out how and if technologies like drones can help control escalation. And what I find in these war games is that technologies like drones, which we often think of as destabilizing because they make it easier and cheaper for states to launch military operations, could actually have a stabilizing effect in controlling escalatory spirals. So to do this using war games, I recruit a bunch of military decision makers and military officers and expose them to a set of hypothetical crisis scenarios. So one of the scenarios that I expose them to is the shoot down of a U.S. reconnaissance aircraft in international airspace. But I randomly vary whether those teams are told that the aircraft is a drone or a manned aircraft. And what I find, unsurprisingly, is that on average, the teams that are exposed to a manned aircraft are much more likely to recommend escalatory retaliation against the adversary. They're willing to violate the sovereignty of another state, to go retrieve down personnel, compared to those that lose a drone. But what I think is really unique about these war games compared to other social science, international relations, research methods, surveys, is that you're really seeing kind of the interactive nature of the decision making process. You can see the debates that happen between practitioners and how they settle on these decisions in a way that I think is really enlightening, especially in cases where we don't have all that much openly available empirical evidence.

John Gans [00:15:32] That's amazing. So how do you factor in A.I.? Artificial intelligence is one of the biggest tech advances out there. Your research has looked at the use of A.I. in the military sphere. You just published a great article in the Texas National Security Review, which I encourage all our listeners to give a read. What are the biggest challenges to the US military and other militaries in using A.I. effectively?

Erik Lin-Greenberg [00:15:59] Sure. So I think states face a set of challenges when they integrate any type of technology. And this is something that Mike Horowitz, our director, really studies in his book, "The Diffusion of Military Power." But I think with A.I., there's a very unique set of vexing challenges that makes it particularly difficult to integrate. And so I think there's, first off, a set of ethical considerations and challenges associated with delegating tasks and decision making to machines. So, for instance, who's responsible if in an A.I. enabled system, something goes awry? Is it the commander of the unit that's operating it? Is it someone else? And there's also questions about in wartime situations that are plagued with fog and friction, whether these systems are as accurate as human decision makers. And so there are D.O.D. policies and policies within allied nations that try to guide these. But I think the public and policymakers are still really justifiably concerned about delegating these decisions. So I think that's one set of ethical considerations. But then there's also, I think, a set of really operationally focused considerations. So A.I. enabled technologies require just this massive amount of data in order to operate and to train them. And so much of this data, though, is really stored in non-standardized formats and in different locations. So a few months ago, Lieutenant General Jack Shanahan, who's the director of the Pentagon's Joint A.I. Center, came out publicly and commented that this lack of data discoverability and standardization really hampers the development of A.I. by the U.S. military. So I think that's one challenge that the military will have to overcome. And then when we think about how militaries operate today, it's often in these coalition contexts operating alongside allies and partners. And so we'll need to find a way not only to share information in the U.S. military, but more broadly with, again, allies and partners. And then third, A.I. really has the ability to speed up the tempo with which military operations take place. And I think this can potentially cause challenges for decision makers who are used to maybe having a little bit more time to make decisions. And this is all even before we consider ways in which adversaries might use A.I. in a various ways. So all these things together, I think, pose a set of difficult political and operational challenges that make A.I. integration quite difficult.

John Gans [00:18:32] If we can transition to what we call the "Franklin Few" here, these are questions that Ben Franklin, one of Penn's first trustees, developed over 300 years ago. And of course, Ben Franklin knew a thing or two about technology and fighting wars. And he developed a questionnaire to be used for conversations among fellow Philadelphians interested in current and global affairs. We've updated it for use today and to anchor our Global Cable podcast.

John Gans [00:19:01] So these are just a few short questions that can have short answers or long answers depending what your mood is. And so the first question we always ask, which tends to be among the harder ones we ask, is who would you most like to meet today and why?

Erik Lin-Greenberg [00:19:15] So I think I'd like to meet this relatively obscure figure from U.S. history, Edward Lansdale. He's a fellow Air Force officer, he's a major general that serves really, I think, on the frontlines of many of the U.S.'s Cold War battles. And he worked extensively with the intelligence community on counterinsurgency operations and was really considered to be a specialist in what we might now call "gray zone conflict" and using information operations and also countering adversary disinformation campaigns. So I think he might have a lot of very interesting insights that would be beneficial given our current security environment and I'd love the opportunity to sit down and chat with them over a coffee or a whiskey or something.

John Gans [00:20:01] The next question is, and everybody under quarantine is probably reading a little more or watching a little more, listening more than they have in the past. Have you read anything or seen anything or listened to anything related to world affairs that our listeners might be interested in?

Erik Lin-Greenberg [00:20:17] Yes. So one of the books on my quarantine reading list is a book by Michael Desch, a professor of political science at Notre Dame called The Cult of the Irrelevant. And the book really explores the role that social scientists have played or not played in forming national security policy within the United States over the course of the 20th century, and I think it offers some really interesting insights, both for scholars and policymakers, on how to do and use political science research in the national security domain.

John Gans [00:20:54] So do you know of any individuals in the United States or elsewhere who have recently done something that deserves praise or imitation?

Erik Lin-Greenberg [00:21:02] So I think this is really a group of individuals, if I can, that often go really unnoticed, but are playing a very important role today as we go through this unprecedented global health crisis. And these folks are helping us maintain at least a sense of normalcy. And so these are the individuals that are our delivery food drivers, truck drivers, grocery store employees, that make sure Philadelphians and the rest of us throughout the United States are really able to get groceries and try to carry out our lives in as normal as a sense as possible.

John Gans [00:21:40] That sounds okay. And then the last question is, is there anything Penn and Penn students can do to be of service to the country or the world today?

Erik Lin-Greenberg [00:21:49] So I think this might seem rather straightforward. But again, given the current crisis, I think it's just to really look out for others. And there's a lot of uncertainty in the world, so just checking in on friends and family and those that might need help I think can make really a world of difference today.

John Gans [00:22:09] That sounds good. Well, Erik Lin-Greenberg, thank you so much. Thanks for joining us here on the Global Cable and I look forward to seeing you again soon.

Erik Lin-Greenberg [00:22:17] Thanks so much, John.