"The problem is you don’t have enough bandwidth,” Schmidt said. “You’ve got to get the networks upgraded. ... you don’t have enough software people," who though "often obnoxious ... can change the world." Mike Tsukamoto/staff
Photo Caption & Credits

USAF’S Big Goals Meet Tech Challenges

March 23, 2022

The Space Force wants to build the nation’s first “digital service.” The Air Force wants to embrace cutting-edge technologies like artificial intelligence (AI), integrated networks, and digital engineering.

Yet to achieve these lofty goals, the services, along with the rest of the Pentagon, need to make some fundamental changes, says the former Google CEO—changes that would put them more in line with Silicon Valley than Washington, D.C.

Eric Schmidt served as CEO and executive chairman of Google and its parent company Alphabet from 2001 to 2017, overseeing a company with tens of thousands of employees and tens of billions of dollars in revenue. He also worked as the first-ever chairman of the Defense Innovation Board and the chair of the National Security Commission on Artificial Intelligence. His own personal net worth, Bloomberg estimates, is around $25 billion.

The next battles will be fought based on software supremacy.

Eric Schmidt, former CEO, Alphabet

In a March 3 keynote address at the AFA Warfare Symposium (AWS)  in Orlando, Fla., Schimdt promised transformative impacts from AI and other software. But he also offered a blunt assessment of where the Defense Department stands today on those fronts—and what it needs to do better.

“In the tradition of the military, I will be direct and I hope that’s OK,” Schmidt told an audience of Airmen, Guardians, and industry leaders. “If I look at the totality of what you’re doing, you’re doing a very good job of making things that you currently have better, over and over and over again. 

“But I’m an innovator. And I would criticize, if I could say right up front, that the current structure, which is an interlock between the White House, Congress, the Secretary of Defense’s [office], the various military contractors, the various services and so forth, is a bureaucracy in and of its own. And it’s doing a good job at what it has been asked to do, but it hasn’t been asked to do some new things.”

From there, Schmidt had plenty of suggestions of new things to try.

First, Fix the Networks

Perhaps the most basic, fundamental problem Schmidt addressed is one that plenty of Airmen and Guardians likely already knew.

“The real problem you have is that you don’t have enough bandwidth … no one ever tells you this,” Schmidt said. “Your networks, excuse the term, suck. You’ve got to get the networks upgraded. And you just have to, because all of these things depend on that kind of connectivity, right?”

Those comments were met with warm applause from his audience full of service members—the issue of poor network connectivity and IT systems is a constant source of frustration among Airmen and Guardians.

That audience reacted similarly when Chief Master Sergeant of the Air Force JoAnne S. Bass brought the topic up during her own AWS panel discussion and acknowledged she is also frustrated “beyond belief” by how slow, cumbersome, and sometimes outright unusable the IT systems can be.

“Our Airmen always say, ‘I wonder if our leaders know, I wonder if our leaders understand the challenges we have.’ And I’m like, ‘Yes, we do, and we share those challenges, right?’” Bass said. “As many times as you have to add in your PIN, I have to do that too. I mean, I send stuff home to my phone or my whatever so that I can actually watch whatever I need to watch, because I can’t do it on my work [computer].”

Without better networks, most of Schmidt’s suggestions become far less feasible.

Staff Up

There are more than 600 projects currently underway in the DOD devoted to artificial intelligence, and countless others focused on software and IT modernization.

But as things stand, the numbers simply aren’t adding up, Schmidt said, whether it’s in staffing levels or funding. Schmidt’s comments echo the final report from the National Security Commission on Artificial Intelligence, which called for an increase in funding from $1.5 billion a year to $8 billion.

“Every time you try to do something in software, one of these strange scavenging groups within the administration takes your money away. It’s insane,” Schmidt stated. “The core issue here in the military is you don’t have enough software people. And by software people, I mean people who think the way I do. You come out of a different background, and you just don’t have enough of these people.”

The software people the DOD needs, Schmidt acknowledged, aren’t like those many typically associate with the military—“they’re often very obnoxious, … they’re difficult, they’re sort of full of things”—but they’re necessary because “they can change the world,” he suggested. 

This issue is particularly glaring when it comes to AI, Schmidt said. And not only does the military lack the necessary personnel, the defense industrial base does too—touring the symposium’s exhibition hall, Schmidt said he only saw “like two AI companies … and by the way, they’re the little ones in the corner.”

Again, Schmidt’s comments echo the commission report, where he and his vice chair wrote that they “worry that only a few big companies and powerful states will have the resources to make the biggest AI breakthroughs.”

There have been some examples of breakthroughs, Schmidt said, pointing to Project Maven, an Air Force AI project that awarded a contract to Google but ignited controversy among its employees in 2018. Nearly four years later, Schmidt said Maven has had “very successful classified use in the right ways.”

“That’s an example of something where you pick it and you fund it, and you weigh it and you build it, and you build the constituencies and then you have it,” Schmidt insisted. 

But Maven was just one project, and “to be very blunt, you don’t have enough people, you don’t have the right contractors, and you don’t have the right strategy to fill in this,” Schmidt said of the Pentagon’s work in AI. “We need 20, 30, 40 such groups, more, more, more. And as that transformation happens, the people who work for you, the incredibly courageous people, will have so much more powerful tools.”

Eric Schmidt, former Alphabet CEO, speaks with AFA’s President Gen. Bruce “Orville” Wright at the Warfare Symposium. Mike Tsukamoto/staff

Corners of Innovation

Yet despite all this, perhaps the biggest issue facing the Air Force’s software and AI efforts isn’t really about software.

“We love to talk about strategy, and we need more money over here—and by the way, we do—and we need more partnerships over here—and yes, we do—and we need more of this over here, and every state has to have its money and all of that’s fine,” Schmidt said. “But what we don’t have and we need a lot more of is the kind of talent to drive this world.”

It’s one thing to say the Pentagon needs to hire more personnel to work on AI and software. In order to retain and grow that talent, Schmidt said, the military needs to empower innovators instead of holding them back, granting them a certain level of autonomy to make decisions and take risks.

Air Force Chief of Staff Gen. Charles Q. Brown Jr. has made empowering Airmen one of the key themes and goals of his tenure. Yet given the massive bureaucracy Schmidt described at the top of his speech— the White House, Congress, the Pentagon, the services, and the defense industrial base—sweeping, comprehensive changes are unlikely.

“When you have such a large bureaucratic problem with so many different stakeholders, you’re not fundamentally going to fix it as an architecture,” Schmidt pointed out. “You’re going to have to adjust it, you’re going to have to make corners, and one of the principles of decentralized leadership is to allow corners of innovation.”

The Air Force and the Secretary of Defense should determine the highest priority areas, Schmidt said, and allow innovators to experiment and pursue cutting-edge technologies.

In that regard, the service could look to its past for a model.

“The Air Force should properly be proud of the Skunk Works model,” he said. And the Skunk Works model in the ’60s was interesting because it was run by a set of colonels, right? And I don’t exactly understand how politically they managed to get the freedom … but somehow they managed to do it on a cycle that was a yearly cycle rather than a 10-year cycle.”

Empowering tech innovators will also have the benefit of helping the military attract and retain talent.

“If I can just be incredibly blunt, you’ve got to figure out that the people that do stuff that I do are like doctors, in the sense that they’re specialists, and they want to be doctors, right?” Schmidt said. “The military doesn’t take these, again, beautifully trained medical people, you don’t just transfer them out to other activities. You have a career path. And they’ll stay … because they believe in your mission. They believe in you. They believe in your culture. It’s not about compensation. Everyone’s obsessed about compensation, which is always an issue. People want to serve.”

Schmidt’s warning comes just a few months after Nicolas M. Chaillan, the first-ever chief software officer of the Air Force, abruptly announced his resignation in a candid LinkedIn post expressing frustrations about the Pentagon’s bureaucracy and lack of appreciation for “prioritizing IT basic issues.”

There is an inherent tension between the Pentagon bureaucracy and the innovation Schmidt envisions. He recounted an incident while visiting Al Udeid Air Base in Qatar and watching a USAF general authorize a strike against a target after consulting with his lawyer.

“What I learned from that was, you’re going to have to have top-down control when kinetic force is used. But you’re also going to have to find a way to give autonomy to the person to do it,” he  said. “And that is at the root of your cultural problem. You want the centralization for protection of the institution, for good reasons. But I also want the autonomy for our men and women to do what they need to do and do it quickly and well.”

Why It’s Needed

Schimdt’s intense enthusiasm for software development in DOD is born out of his belief that “the next battles will be fought based on software supremacy. And they really will be. And you understand this, you’ve heard it, but you don’t have it yet.”

In particular, “AI is a force multiplier like you’ve never seen before,” he said. “It sees patterns that no human can see. And all interesting future military decisions will have as part of that an AI assistant.”

Schmidt is hardly alone in predicting AI will have a seismic impact on warfare. It has the capacity to transform missile defense, precision weapons targeting, precision analysis, and autonomous systems.

But advocates say they’ve also encountered resistance and inertia within the Pentagon, preventing the U.S. military from fully embracing AI’s possibilities and forcing service members to “spend all day looking at screens doing something that a computer should do,” Schmidt said.

As the Air Force and Space Force look to transform themselves into hotbeds of innovation, there is one current internal example they can look to for inspiration, Schmidt added—the B-21 Raider program. Praising the Rapid Capabilities Office, Schmidt said the Air Force developed the new stealth bomber in a “new and innovative” way. 

“Think about the B-21 example, but apply it to things other than bombers,” Schmidt urged. “Like, let’s try to do the same thing for software.”