The U.S. has the largest defense investment infrastructure in the world. It’s a highly complex, intelligent, and agile environment driven by a historical and global imperative to protect democracy.
Concepts such as OODA (observe, orient, decide, act) and VUCA (volatile, uncertain, complex, ambiguous) have reshaped the combat mindset in transformative ways. In a world where digital-everything increasingly dominates thinking, the need to transform behaviors to focus digitally is as important — perhaps more important — than simply transforming in the physical world. Cyberthreats, the use of AI, cloud re-tasking and cross-functional real-time collaboration are all part of the new intelligent systems world.
The situation is complex, global in nature, and increasingly seen as the Gordian knot at the center of global defense for the two most advanced military nations on the planet. One (China) has aggressively grasped that today’s world centers around digital dominance. The other (the U.S.) has yet to fully grasp the magnitude of the changes, threats, and opportunities that AI, and related concepts, represent in this new intelligent-systems world.
We interviewed one of the key players in the drive toward raising awareness of the threat — and toward creating a more digital way, such as the U.S. Air Force’s Platform One, that will counter the illusion of permanence created by military tradition. Nicolas Chaillan was the first chief software officer (CSO) for the U.S. Air Force. His ideas about DevSecOps and a single, integrated delivery system have been thoroughly documented elsewhere (Nextgov.com). This interview takes the discussion into the multidimensional nature of change the U.S. faces in a cyberthreat-led world where AI and automation drive consistent competitive advantages.
Why, at this moment, are we moving away from hardware-only to intelligent systems defense?
Chaillan: There is a tendency for large institutions to ignore the obvious, until someone doesn’t just raise a flag but raises lots of flags and says, “Hold it, something’s wrong here!” I think you have an interesting fulcrum of software for that new digital warfare space that’s inevitable, which we seem to be ignoring in this country.
I think the future is going to be very different from what we’ve been preparing for. I think the U.S. government has been investing massively in hardware innovations and very little in software, cyber offense, cyber defense, and artificial intelligence. You see the U.S. government spending a lot of money on fifth-generation fighters and war-fighting capabilities that are, quite honestly, probably now seen as old ways of fighting wars. I think you might realize pretty quickly that the next wars will be fought through a lot of ruthless automation, based on the ability to continuously update and deliver software and react to offense and defense issues.
You’re going to see, for example, the rise of AI hackers — robots that will be used to get into U.S. systems. You’re going to see AI used to make better, faster decisions and be able to compound information and gather more intelligence. (You see it with the adoption of TikTok, which is effectively just that.) I think that’s going to be a very interesting and game-changing look at what we’re going to be fighting in the next 20 years.
Can we talk about the conflict of trying to fight the next war using the mechanisms of the old war, and why we struggle to migrate to this new software, AI, autonomous, highly agile environment versus the traditional large blocks of pure hardware?
Chaillan: Just as Silicon Valley is a bubble, you also see a bubble of talent in the DOD. What you see is people who are not coming from different backgrounds. Then, even when they leave the government, they end up working for some of the defense industrial base companies that are also part of the bubble. We don’t invest in our people and we don’t proactively drive continuous learning for better understanding of the importance of software, which is not just a “nice to have” but a core new competency.
We are starting to see a lot of near-real-time software innovations, such as what SpaceX is able to do, including updating software the day before the launch of a rocket. And they can reuse rockets, which is the de facto standard for any kind of new innovation. Today no company would try to build a launching capability that doesn’t reuse rockets — that was not the case even five years ago.
How do we break that Gordian knot? Because at some point, trying to do the same thing and get different results is not likely to succeed.
Chaillan: We need to be able to find a way to bring talent into the government from industries that are outside of the traditional duty bubble. The other piece is investing in our people. We need to empower, then get out of the way of the talented people in our military.
We also must invest in their continuous learning. I was giving an hour a day to my people at Platform One to learn. This was essential so they could not only catch up but then keep up with the crazy pace of it. We need to do better there, and continuous learning will be the difference between becoming stale or leading.
Why does the intelligent systems world (cloud, AI, edge) force the need for change?
Chaillan: We need to find a way to convince 150 leaders to work together across all parts of our defense environment. We need to break those silos. We need to stop the Army, Navy and Air Force from building things in a vacuum. We have to deal with cloud and zero trust and DevSecOps and data fabric and AI/ML layers.
We have 58 networks with all-mission partners, including NATO and Five Eyes. And for them to communicate with each other is just impossible. You end up having to copy and burn DVDs between devices.
I see change happening with JADC2, the Joint All-Domain Command and Control program. JADC2 is effectively supposed to be an “internet of things” connecting all weapons into a central capability to get better insights and more effective capabilities, which defeats the point of creating silos. Now you’re paying them to integrate stuff that was designed to be siloed, built in vacuums. They’re addressing issues such as basic networking — anyone would agree that we could have a transport layer bringing best-of-breed fiber-optic commercial internet, plus 5G satellite, plus the cloud service provider backbone into a single mesh, with crypto on top to access different classification levels.
Why is the pace of adoption elsewhere of ideas such as AI and automation so concerning?
Chaillan: We will not be able to catch up to the velocity and pace of China’s adoption of AI, because they have 1.5 billion people and access to data points that we don’t have the luxury to get access to. Add to this the velocity of delivery of these capabilities in production, which compounds over time. AI is going to learn and add more data and more training, with better algorithms, getting stronger over time — compounding upon itself. At some point, you just physically cannot catch up.
You’re going to see a lot of what we do today with humans being completely automated. Take imagery, for example — looking at satellite data and being able to understand what’s going on, having AI continuously monitor what’s going on and give us a better picture of and insight into what’s happening. A human cannot physically keep up with all this imagery, zoom at the right place and do all the stuff we need to do. But AI could potentially do that seamlessly across dozens of capabilities, aggregating data.
You look at Tesla: They open sourced their patents, because they’re moving so fast that by the time other companies even start using them, Tesla’s going to be five miles ahead. We should be in the same situation.
An intelligent systems defense-oriented world has to synthesize the physical and digital worlds’ best attributes (automation, AI/ML) to be able to react and re-task through the cloud.