DoD Has 600 AI Projects–Needs Out of the Dog House

The Pentagon is a very large building that houses a lot of the authorities for the U.S. Military. It's big. This photo doesn't do it justice, but they're kinda rightly concerned about people flying drones around there.

The Pentagon has a lot of dogs in the artificial intelligence fight. Now it seems to be setting up shop to get those puppies groomed and ready for the big show.

“I was told that we have 592 separate AI-related projects across the department,” Michael Griffin, under secretary for research and engineering–essentially Departmnet of Defense (DoD) CTO–told the House Armed Services committee Tuesday. “We need to bring some focus to all of that.”

Griffin confirmed Defense Secretary James Mattis’ remarks a week prior regarding the establishment of an “AI Center.” Griffin hinted at collaboration with intelligence agencies and universities within the newly established office, which will lead research and development, hoping to fine tune those hundreds of efforts. Organizational roles are still in flux, Griffin noted.

“We are looking right now as we speak about things like: how do we structure it; who should lead it; where it should be; how we should structure our other departmental research to focus in,” he said. “These are ongoing questions that we’re addressing this week.”

Leadership within the new department is a pressing concern, but that’s not all that’s on the shortlist of DoD’s imperatives in getting out on emerging technology.

“I am concerned that we are not leveraging our technical advantage,” Griffin stated. Part of that rests in getting qualified help.

“The majority of the contractors that are used by the DoD are not AI-capable at this moment,” said Eric Schmidt, chairman of DoD’s Defense Innovation Board.

Griffin outlined the real-world implications of inaction, when adversaries are developing new weapons that outpace detection systems.

“The most significant advance by our adversaries has been a Chinese development of what is now today a pretty mature system for conventional prompt strike at multi-thousand kilometer ranges,” he said. “We will, with today’s defensive systems, not see these things coming.” Congress hears the sirens wailing.

“The more we can do to cut out the red tape and accelerate these programs, the better off our nation will be,” said Rep. Jim Langevin, D-R.I., ranking member on the Emerging Threats and Capabilities subcommittee. He asked Schmidt what the most imperative recommendation was to spur innovation; Schmidt referenced the AI office first.

“We are specifically proposing that the nature of AI is a long-term technology that will be useful for defensive and perhaps offensive purposes as well,” Schmidt said.

Organizational process is important, but only insomuch as it accelerates development, and results. To paraphrase, all this AI bark needs to be met with the ability to bite back.

Recent