DOD AI chief: 'Grave misperceptions' about military programs

The Defense Department is no stranger to controversy when it comes to artificial intelligence. Employees at Google and Microsoft have protested their companies’ involvement with DOD, concerned about unaddressed ethical issues around designing technologies that could enhance lethality.

But for Air Force Lt. Gen. John "Jack" Shanahan, the Defense Department's Joint Artificial Intelligence Center, the much-needed discussion has gotten off track thanks to "grave misperceptions" around what technology DOD is developing and for what purpose.

"There are grave misperceptions about what DOD is actually working on," said Shanahan, who also led the controversial image perception Project Maven with Google, during a keynote address at the AFCEA AI and machine learning summit March 27.

"In my experience in the last few years, what I've found is there is an assumption, in some quarters, that the DOD is in a back laboratory somewhere in the basement of a building, has got a free-will agent AGI -- artificial general intelligence -- that's going to roam indiscriminately across the battlefield. We do not."

Shanahan said artificial intelligence is a tool used for specific problem sets, and like every other technology used by the DOD, will be evaluated for legal, ethical and moral concerns.

The JAIC is focused on four capability areas: intelligence and perception (think Project Maven); predictive maintenance with an Army pilot looking at mechanical issues, maintenance, performance and personnel management; disaster relief and humanitarian aid; and cyberspace. In fiscal 2020, the JAIC will also work on an effort targeting peer competitors and the "full spectrum of DOD operations."

Shanahan wouldn't discuss the effort beyond it being tied to the National Defense Strategy and noting that "this is so important, it's potentially so big, that we're going to spend more time on the problem-framing part of this so when we get our funding and people in fiscal year '20 we can accelerate."

Shanahan made similar points in recent testimony before the Senate Armed Services Subcommittee on Emerging Threats and Capabilities, pushing the JAIC's role and its partnership with the Defense Innovation Board, which had its first public meeting earlier this month.

"To underscore our focus on ethics, humanitarian considerations, and both short-term and long-term AI safety, JAIC is working closely with the Defense Innovation Board (DIB) to foster a broad dialogue and provide input into the development of AI principles for defense," Shanahan said, according to written testimony for that March 12 hearing.

During the question and answer portion of his March 27 keynote, Shanahan said that AI policy for the legal, moral and ethical concerns around AI exists, but there will need to be an element of transparency to get conversations back on track.

"Accountability and transparency matter but somehow that conversation got off-track with some aspects of industry largely because of the assumption of what DOD might be working on then what we're actually working on," he said.

"We know there's work to do to continue a healthy dialogue about what our value system is, how we do adhere to international norms and how some of our potential adversaries are likely not to," Shanahan said. And while these discussions are important, he said, it "doesn't hold us back from moving forward with AI across the full range of DOD missions."