3 December 2018

Army AI Task Force Comes To Pittsburgh, c/o CMU

By SYDNEY J. FREEDBERG JR.

DETROIT: The Army’s brand-new Artificial Intelligence Task Force — just launched last month — has set up its main office on the Pittsburg campus of Carnegie Mellon University, Gen. John “Mike” Murray announced here this morning. The task force’s deputy commander, a colonel, is already working out of CMU’s National Robotics Engineering Center, I was told. The task force will use CMU as a hub to reach out to academia and civilian industry, Murray said, as well as coordinating existing Army AI efforts, starting new experiments, and reforming bureaucratic policies that impede innovation.

With Murray’s own HQ, Army Futures Command, stood up this August in Austin, the notoriously hidebound Army is making a concerted effort to change its ways. Reaching out beyond its comfort zone of the Pentagon, big military installations, and traditional defense contractors, the Army hopes to embed its officers, scientists, and civil servants in a wider world of civilian innovation, where hoodies are more in vogue than uniforms.


Carnegie Mellon in particular has a long history with the military, having developed the most successful robot (a modified Humvee) for the DARPA Grand Challenge and a robotic mini-tank for the Army’s cancelled Future Combat Systems. But CMU also has a sterling reputation in civilian circles: It has a US News rating as the nation’s No. 1 graduate program in computer science and AI, as well as strong ties to both other universities and leading companies.

So besides being a valuable source of AI and robotics expertise itself, CMU can help connect the Army to a much wider world that it has struggled to connect with — a world where many key players, such as Google, have outright refused to work directly with the military. In fact, while CMU will be the starting point for the Army AI Task Force, Gen. Murray said, “from that initial hub, the AI Task Force will extend a network across the academic and industrial ecosystems.”

The Army’s still working on what form that would take, the AI Task Force director, Brig. Gen. Matthew Easeley, told me on the sidelines of the AUSA AI conference here. One option he’s considering is a network of satellite offices similar to those set up by the Defense Innovation Unit (DIU, formerly DIUx).


BAE’s Armed Robotic Combat Vehicle (ARCV), originally developed for the Future Combat Systems (FCS) program. Carnegie Mellon developed the navigation system and other key components, although they didn’t work on the armaments.

Don’t Spread Too Thin


At the same time, Easeley emphasized, the Army doesn’t want to spread itself too thin, as it often has in the past. “We didn’t want to just continue our normal process of having a little bit of money scattered across a large number of universities, so we’re trying to get a focused effort at a few key universities around the country,” he told me.

One potential roadblock Easeley didn’t address: Congress likes spreading federal funding out across the country. The Army will have to convince both authorizing and appropriating committees that it’s better to focus their investments — and hope legislators whose districts do get funding can out-hustle any sore losers on the Hill.

Matthew Easeley being sworn in as a brigadier (one-star) general in January 2018.

Part of the reason for concentrating the Army’s fire is that it just doesn’t have a lot of resources to spread around. Between President Trump’s surprise $33 billion cutto national security spending and the Democratic takeover of the House, the Defense Department budget is under pressure — although Deputy Secretary Patrick Shanahan has vowed to protect Army modernization.

Even if resources were abundant, there’s also a case for keeping the Army’s footprint light and unobstrusive, rather than trying to flood the zone with soldiers and bureaucrats — which would probably trigger a violent immune reaction from most techies.

“Our task force will be small,” Brig. Gen. Easeley emphasized: “We expect ti to be less than 20” — about two-thirds uniformed military and one-third government civilians, a mix of warriors, engineers, and budgeteers. They’ll be split between the main office in Pittsburgh, run by Easeley’s deputy, and an office in the Washington area, which is where Easeley himself will be to better connect with Pentagon leadership.

But CMU will clearly be the hub. “I know a number of the senior army leadership has either been there or is going to go there, (and) the Air Force leadership is doing a similar pilgrimage next week,” Easely told me “We’re hoping to get CFT leads” — the eight Cross Functional Team directors working on the Army’s top priorities — “there periodically to help them understand how AI can help their specific areas,” from long-range missiles to robot tanks and infantry gear.

Of course, the Army’s tried to revolutionize itself before, with distinctly mixed results. What makes this time different?

Easeley’s answer: It’s that the Army has broken open its organization for the first time in 40 years and created a new top-level command, led by a four-star general, that’s specifically dedicated to modernization and which brings together expertise to do it that was formerly scattered across the bureaucracy.

“By creating the four-star headquarters, that dramatically changes the way we’re doing this,” Easeley told me, “especially since we’re bringing the requirements piece, the engineering piece, the science and technology piece, all into the same organizational structure. (That) just didn’t happen before.”

Gen. John Murray, first chief of Army Futures Command, speaks at its formal activation in Austin.

Gen. Murray’s Remarks

Since Gen. Murray is the four-star in question, it’s worth reproducing his remarks this morning at some length (my transcription, checked against audio):

“Many of the military artificial intelligence solutions that uniformed leaders are so excited about are just new applications of AI tools that have been in the hands of the private sector and private citizens for the last decade or so.

“To focus our efforts in this constantly changing field, the Army established an AI task force operating primarily at Carnegie Mellon University’s National Robotics Engineering Center in Pittsburgh, Pennsylvania. From that initial hub, the AI Task Force will extend a network across the academic and industrial ecosystems.”

“As of this morning, the AI Task Force is focused on completing their initial setup and planning their way ahead. In the coming months, they will identify existing and ongoing machine learning initiatives within the Army, as well a framework for implementing small projects that can evolve or combine into something much larger.

“The Task Force will establish an Army AI testbed and will review policies that could inhibit machine learning, deep, learning, and automation, and perhaps most importantly … they will develop a talent management plan to enable the Army to hire, develop, and retain the necessary skill sets to support AI development well into the future.”

No comments: