Capt. Mike Kanaan woke up at 3 A.M. with an idea. The 29-year-old U.S. Air Force intelligence officer had been mulling a familiar challenge, the problem of finding coders and computer-savvy airmen, who are in high demand and short supply all across the Pentagon. His wee-hour solution: Treat computer coding skills like the service does any other mission-critical foreign language.
The Air Force prides itself on its global reach, and so the service fosters and rewards foreign language skills. The Air Force measures linguistic aptitude with tests to determine an airmen's existing fluency, or his or her capacity to learn another language. Those who pass are open to duty assignments that require another language—and get extra pay as well.
“The languages of computers themselves are as critically significant to the needs to the DoD as much as any of these traditionally viewed, non-native languages,” Kanaan says. “Python, Java C++. They all have their unique vocabulary, and sets of grammatical and construct rules, just like any other language.”
The idea is simple. It's also disruptive. Harnessing coder talent would create a cadre of known airmen ready to face the digital battlefields of the future, and the idea has taken root. In less than two months, Air Force intelligence HQ launched a pilot program that is being examined for wider, Pentagon-wide adoption.
You Need People
Kanaan says these computer language skills are vital for weathering the current sea change in battlefield analytics, big data, and artificial intelligence. He ought to know: He serves as the enterprise lead for artificial intelligence and machine learning at U.S. Air Force Intelligence headquarters.
“There’s a lot that goes behind delivering any meaningful application of AI at scale. You have to have the right level of compute; you have to have the appropriate data,” Kanaan says. “But another key piece is human-centric. You need people. People make algorithms.”
There is an AI arms race going on right now. If you’re thinking Terminator versus Terminator, stop. This struggle is about ways to use enormous data sets to make a military more efficient, and therefore more dangerous.
Competition from near peers, especially China and Russia, has given the Pentagon pause. Misinformation campaigns pushed out by ‘bots, increasing use of satellites and drones, and persistent cyberattacks pose unique threats to the United States. “We are trying to get back and sustain our competitive advantage,” Kanaan says.
The Pentagon fears it is behind the curve, and is trying to remedy that. Last week, DARPA announced $2 billion in investments over the next five years flowing into new artificial intelligence research. This summer also saw the birth of the Joint Artificial Intelligence Center. “This is just one pillar to delivering meaningful analytics and meaningful artificial intelligence,” he says of his drive to put coding on par with other languages.
Kanaan is now a rising star within the Pentagon. During , he led 1,800 Air Component Commander analysts to create the first-ever visualization of millions of data points from sources including unmanned aerial vehicles and open-source streams. That reduced 1.5 hours of daily target development to five minutes. He also provided intelligence support to the fight against ISIS that increased deliberate (i.e. pre-planned) target development by 68 percent. Now he’ll be moving up again, soon to become the co-chair for Artificial Intelligence for the Air Force.
The Hidden Experts
Killer robots, and whether machines should decide when or if to shoot, are the focus of much angst right now. But behind the scenes, machine learning can help analysts find targets in cluttered landscapes, cut maintenance costs by automatically diagnosing emerging problems and ordering replacements, validate communications that could be tampered or hacked, and even making bean counting more effective by sniffing out duplicative efforts among reams of financial data.
AI will change the military in some fundamental but unseen ways. Kanaan says the preferred way to sell leadership on artificial intelligence is to prove it in real world ways—what he calls “buying yourself into existence.” The right application, using the right level of machine learning to spot patterns and adapt to changes, can create a believer.
For example, when Defense Innovation Unit went to air operations centers in Middle East, the defense tech expert envisioned software changes that would optimize the way that airmen tracked refueling tankers. Teaming up with a commercial firm in Boston called Pivotal Labs, the new software is saving about $200,000 in fuel every month.
Other advanced software examples are cloaked behind operational secrecy. “Within Air Force intelligence, we have a DevOps environment where we have a ton of high performing computes at the classified level,” Kanaan says. “I can’t say what they’re doing but we’ve created that environment to build those applications.”
Computer language is a key part of AI adoption, and these skills may be lurking among airmen who are not in traditional roles. Kanaan is looking for those with hacker mentalities and those drawn to digital tinkering. “Statistically speaking, some of the most talented programmers in the Department of Defense are not going to be in places like cyber or intelligence, they are probably going to be a Bitcoin miner at home or just someone with a passion."
The broad focus makes it a complement to U.S. Cyber Command, not a competitor, Kanaan says. “We’re super-close to Cyber Command,” he says. “It’s a mutually beneficial process. Not only will I bring a same capability that is indigenous to them, but I’m bringing that out to the rest of the force. And in addition, we’re going to now identify people who need to put those skills sets into Cyber Command. It’s a Yin and a Yang thing here.”
One of the things that the pilot programs must prove is that the coding and programming skills will be put to good use. Success would be defined by a wider rollout within the National Defense Authorization for 2020. Kanaan says he’s ready for that challenge and has no doubt airmen will prove the case. “We will have to start building these skillsets into everything. ”