‘Constant stare’: US Pacific commander wants AI to tell Chinese military exercises from invasion

WASHINGTON — Record surges of Chinese warplanes over the Taiwan Strait. Quasi-civilian Maritime Militia vessels, over a hundred of them, “swarming” in Philippine waters. The Chinese Coast Guard blasting water cannons at their Filipino counterparts. China’s first domestically built aircraft carrier leading a score of warships in wargames of unprecedented scale. All that in just the last six and a half months.

So as Beijing keeps raising the temperature in the West Pacfic — like the old myth of someone boiling a frog alive, too gradually for the amphibian to notice — how do intelligence analysts determine what’s just one more maritime micro-aggression, and what’s the opening moves of a Pearl Harbor-style first strike?

They should use artificial intelligence, and American industry needs to built those analytic algorithms ASAP, according to Biden’s nominee to helm US Indio-Pacific Command.

“Just in the last three years, [we’ve seen] step-level changes in the force levels, the jointness of those force levels, the rehearsals,” Adm. Samuel Paparo told an audience at a Silicon Valley conference hosted by the Defense Innovation Unit. “[It’s] raising the threshold of warning… Soon we’ll be at a point where a force sufficient to execute a profound military operation is in the field and operating under a fig-leaf of exercise.”

RELATED: Pacific Fleet chief Paparo on China’s big lesson from Ukraine: Win quickly

It’s not a new idea that hostile forces might move into position to attack on the pretense of conducting training exercises. Both the US and Soviet Union kept an anxious watch for such strategems during the Cold War, for example, to the point that a top-level KGB team called Project RYaN panicked that NATO’s Able Archer/Reforger wargames in 1983 were actually cover for a nuclear first strike. Ultimately, however, neither NATO nor the Russians ever pulled the trigger because they mistook wargames for the real thing, nor did either side ever get blindsided by a surprise attack.

But longstanding intelligence methods, used to suss out “indications and warning” of a first strike in time to brace for impact, are no longer adequate to the speed and complexity of the Western Pacific today, Paparo said.

“This erosion of strategic, operational and tactical warning is real,” he told the conference. “[It] presents a challenge to the joint force in our ability to go deeper, to find those indications and warnings that will enable us to be postured to support our allies and partners, and — if so called by the commander in chief … defend Taiwan.”

“When we are left in a spot where we can no longer derive enemy intentions — would-be enemy intentions —from their disposition in the field, we must go deeper and that requires data, compute, talent,” he said. “What are those indications and warnings?”

That’s a geostrategic and political problem, not just a matter of military operations, Paparo emphasized. “The ability on short notice to see, understand, decide, and act against a rapid effort on the part of another great power … allows decision makers, the commander-in-chief and the Congress maximum decision time and space to to act when there is no doubt that there is violence impending.”

But that’s lacking today, he said: “That ability to see and sense what our competitors are doing so we can be better prepared for aggression, I think, is is an important gap.”

The commissioning ceremony of the Shandong aircraft carrier is held at a naval port in Sanya, south China’s Hainan Province, Dec. 17, 2019. (Xinhua/Li Gang via Getty Images)

Constant Drone Stare, Instant AI Analysis

Advanced technology can help bridge that gap, Paparo told the tech industry audience at the conference in Mountain View, Ca. Specifically, he wants a large number of low-cost, long-endurance unmanned sensor platforms — drones, robot ships, and so on — to keep a closer watch on Chinese activity than manned scouts ever could.

Then he wants to feed the resulting firehose of data, far more than human analysts could ever sort through, into artificial intelligence algorithms that can detect anomalies and patterns quicky and accurately enough to give commanders warning.

“The proliferation of sensors gives us a stare instead of a blink,” Paparo said. “We can employ swarming technology [with] machine learning on board … to have a constant stare on the operational environment, that is shareable throughout the joint force and allied force, to be able to see an adversary’s behavior.”

That “constant stare” of surveillance data then must feed into new kinds of “mass-data analytics … and processing, evaluation, dissemination of intelligence products to go deeper, to be able to determine strategic, operational, and tactical warning,” Paparo said. “[Analysis] that takes human beings hours, if not days, to do can be done in seconds by a machine.”

Backstopping this AI-powered intelligence apparatus should be expendable or “attritable” swarms of unmanned weapons, like those being developed by the Pentagon’s Project Replicator, to blunt any Chinese offensive. The goal, Paparo said, is “unmanned, autonomous capability that – with human accountability – can deny the Taiwan Strait as an avenue of invasion,” as well as Chinese land grabs in the disputed waters of the South Pacific.

RELATED: DIU eyeing Feb-Aug 2025 to field first Replicator systems, wants industry input

This capability is primarily a matter of software, he told the Silicon Valley audience, not hardware.

“In many cases our problems aren’t seekers or propulsion or mass, although I would like to see a deeper magazine of long-range weapons, but how to optimize those tools,” he said. “That’s really not necessarily at the compute level, although there’s a lot of great work on compute and on chip design and all that. … But what applications can you build?”

“It ain’t all just about fires, because as important as delivering warheads on targets, equally important is delivering Class One, Class Three, Class Five, Class Eight supplies to the battlefield,” Paparo said. The goal is to “deliver a true, accurate picture to the warfighter, with instantaneous communications …. with a low probability of intercept, and with the kind of encryption that ensures it’s secure, to unlock human creativity at the tactical edge.”