Robots are Real, but AI’s Full Promise is Still on the Horizon

NATIONAL HARBOR, Md. — Artificial intelligence in all its forms, from machine learning algorithms to unmanned systems, is a sure thing for the sea services and its partners, but there is still much to determine in terms of the technological and operational challenges it presents for warfighting.

In a panel discussion on May 8 at Sea-Air-Space, U.S. Coast
Guard Rear Adm. David Dermanelian, assistant commandant for C4IT and commander
of Coast Guard Cyber Command, framed the conversation as a relevant, real-world
issue for the sea services.

“This is not the art of the future. It’s happening today,” Dermanelian
said.

U.S. Marine Corps Brig. Gen. Christian Wortman, vice chief
of naval research, said the Corps has an expansive approach to AI and is
seeking to embed it into everything the service does, including machine
learning to make war more efficient and help make more informed decisions. But
he stressed that users “can’t look at this in isolation,” and the Marines also
need enhanced network capabilities and to use the cloud so algorithms can take
advantage of the data that is harvested.

U.S. Navy Rear Adm. Casey Morton, who was on day three of
his job as program executive officer of Unmanned and Small Combatants, said his
service is “firmly” moving in the direction of adding more unmanned elements to
its assets, from unmanned surface vehicles to unmanned underwater vehicles and
beyond.

“They are going to be a part of our team,” Morton said. “It’s not a matter of if; it’s a matter of when and how fast and how can we get there.”

Right now, he believes the Navy is not yet at human-machine
teaming but is working toward that future where Sailors and machines work
closely together. He cautioned that there are still a lot of unanswered
questions about AI, like what infrastructure it will need, where it will be
based in the fleet, how it will be supported, if it will be forward-deployed
and other policy issues.

“We are at the early stages of this still,” Morton said. “There
are a lot of questions here that are still unanswered.”

“This is not the art of the future. It’s happening today.”

U.S. Coast Guard Rear Adm. David Dermanelian

The U.S. Maritime Administration’s Christopher Walher, who
focuses on the education programs of MARAD’s six state maritime academies, sees
AI as a pedagogical challenge, since sometimes subject matter experts are too
advanced to be excellent teachers, often skipping over critical points that, to
them, appear obvious.

He prefers a “crawl, walk, run” approach to the training
pipeline, where MARAD leverages a training process so AI can manage what it
excels at and humans can focus on their strengths, much like the current relationship
between smartphones and users.

Key for MARAD going forward will be working with other organizations, including a meeting the agency has next month with AI experts so they can share information, versus starting from ground zero on research and development.

“As we talk about crawl, walk, run in the Maritime
Administration, we are the little ship that could,” he said. “We don’t have a
lot of money for R&D.”

Steven Escaravage, senior vice president for the Strategic
Innovation Group at Booz Allen Hamilton, briefly went over his company’s 60
current programs that involve machine learning and robotics, including areas
like sensor data processing, electronic warfare, predictive maintenance and
optimized planning.

Escaravage said the field of AI in the last six to 12 months
has focused on taking what has been written about and researched in the lab and
tried to operationalize those concepts so they can be used in real-world
environments. He said while AI has suffered from being overhyped, there are
some rich capabilities for it today.

“Although today’s capabilities are probably over-extended
and somewhat brittle, what’s going to happen in a matter of months is going to
be real capability that changes pretty much everything we do.”