Wanted: A new strategy for machine intelligence

As machine intelligence makes its way out of the lab and onto the battlefield, experts are calling for a  national strategy that simultaneously advances and harnesses the technology.

The U.S. needs a roadmap for responsible development of machine intelligence because the nation's current Artificial Intelligence Research and Development Strategic Plan  has been overtaken by events, including a growing list of strategies other nations have unveiled over the last year. Most notably, China released an AI development strategy in August 2017  and continues to pour billions into AI research. That was followed weeks later by a pronouncement from Russian President Vladimir Putin that whoever “becomes the leader in this sphere will be the ruler of the world."

The strategic plan released by the Obama administration in 2016  “has kind of gone stale,” said Josh Elliot, Booz Allen Hamilton’s director of machine intelligence and data science solutions who is also the  co-author of a report released earlier this month by the Center for Strategic and International Studies and underwritten by Booz Allen. The report stresses the “hard power” implications of robotics and other forms of automation, particularly as the Defense Department and the Chinese People’s Liberation Army recognize that “the next generation of military technologies will be driven” by machine intelligence.

The report also seeks to address the often-heated rhetoric about the promise and pitfalls of AI. Elliot said it focuses on the overarching concept of machine intelligence, defined as machines augmenting humans to accomplish a specific task. “AI often connotes 'killer robots,'” Elliot explained in an interview, adding that the report seeks to tamp down the hyperbole around the technology.

Among CSIS' recommendations is a federal role in establishing ethical and safety frameworks for implementing machine intelligence.

The report also focuses on the impact of machine intelligence on national security, both in terms of economic competitiveness and how the technology could transform the battlefield. Among the predicted outcomes are a seismic shift from today’s information warfare strategy to what experts call “algorithmic warfare.”

A case in point is Project Maven, the moniker for a fast-tracked Defense Department effort called Algorithmic Warfare Cross-Functional Team.

Project Maven was launched in April 2017 to accelerate DOD’s integration of big data and machine learning into its intelligence operations. The first computer vision algorithms focused on parsing full-motion video were released at the end of last year, and can be updated almost daily, according to Graham Gilmer of Booz Allen’s machine intelligence team.

“We have analysts looking at full-motion video, staring at screens [6 to 11] hours at a time,” Lt. Gen. John Shanahan, DOD’s director of defense intelligence, told an industry conference last November “They’re doing the same thing photographic interpreters were doing in World War II.”

Project Maven aims to “let the machines do what machines do well, and let humans do what only humans can do” -- the cognitive analytical portion of video interpretation, Shanahan added.

Gilmer noted that this early “indications and warnings” application of machine learning means efforts like Project Maven are “not even close to pulling the trigger,” a reference to autonomous weapons that some AI critics fear. “We need to be testing, we need to be prototyping,” he added.

As DOD rolls out algorithms for testing, the CSIS report noted that the rise of algorithmic warfare nevertheless has implications for future weapons while creating new military capabilities. Among them are the integration of machine intelligence into current C4ISR systems.

“Perhaps the most transformative applications of military machine intelligence are in command and control (C2),” the report notes. “MI-enabled C2 could develop entirely novel strategies, anticipate enemy tactics, accelerate intelligence, surveillance, and reconnaissance and help coordinate activities of large numbers of dispersed units acting in tandem.

“As a greater share of decision-making on the battlefield happens at machine speed, human thinkers may be unable to keep up,” the report predicts.

Indeed, a number of machine learning and analytics startups advised by retired military officers are beginning to explore new approaches that extend beyond current predictive analytics. One approach, dubbed “abductive reasoning,” is touted as helping battlefield commanders understand enemy intentions, distinguishing a bluff, for example, from an actual attack.

Approaches like abductive reasoning and scenario-based war-gaming are among the collaborative technologies embraced in the CSIS report. Elliot, the co-author, said military applications should for now focus on narrow, often tedious, analytical tasks rather than true autonomy.

Along with developing ethics and safety standards (Booz Allen expects to shortly release a paper on machine intelligence ethics), Elliot said a broader federal role in both promoting and managing machine intelligence should take the long view. That includes familiar recommendations like funding high-risk, high-payoff research, promoting workforce skills and other market mechanisms like lowering barriers to entry for technology startups. The effort could also expand recent open-source efforts designed to make government data more accessible, he said.

Funding for a government-industry research consortium could be funneled through the Defense Advanced Research Projects Agency, the Defense Innovation Unit Experimental and, for university-based research, the National Science Foundation, Elliot said.

Congress is also weighing in on the need for a national machine intelligence strategy. The House Oversight Committee's information technology subcommittee is holding a series of hearings on how government agencies can adopt “game changing” AI technologies. DARPA and other agency officials echoed calls for R&D investments, broader access to government data and boosting the AI workforce through computer science and STEM education.

A subcommittee hearing in April will focus on establishing guidelines for promoting machine intelligence development while ensuring “we’re using it in the right way,” said Rep. Will Hurd (R-Texas), chairman of the IT panel.

As the pace of development quickens, a consensus is building that the U.S. must lead from the front. “Machine learning has the potential to be transformative,” Elliot stressed. “It’s important that [the U.S.] not fall behind.”

This article was first posted to DefenseSystems, a sibling site to GCN.