Artificial intelligence tools promise, among other things, to make the Pentagon more efficient.
The first annual report on artificial intelligence was released by the White House in February. This long-overdue document covers developments in the year since the American AI Initiative kicked off.
The report said that American research and development investments in AI “are measured not just by the specific dollar total of the financial investments, but also—and perhaps more importantly—in the quality and impact of those investments. These investments are now paying off in everyday lives, in applications from healthcare to financial services, from weather forecasting to transportation, and more.”
The report is a welcome call to action. The U.S. approach to applying AI needs to be full-spectrum at full speed. We must use all applicable AI technologies, including computer vision, natural language processing, deep learning, and reinforcement learning, and apply these to all critical areas of national security, logistics, transportation, public services, and healthcare. We must also invest in developing secure, high-capacity architectures to enable AI technologies. Only with this focus will we begin to quickly gain experience in AI R&D, and in deployment of associated applications at scale. And it is only then that the benefits that accrue from AI, both to citizens and warfighters, will become manifest in a significant way.
There are a number of reasons why the U.S. government has lagged in adopting “exponential technologies.” While the government has made plain it wishes to partner with the private sector, a variety of bureaucratic, regulatory and policy impediments have prevented this partnership from fully forming. This is especially true for smaller, innovative startup companies that are often the crucibles of innovation and out-of-the-box problem solving.
Despite some recent, innovative changes to acquisition processes, the usual bureaucratic minefields await bright-eyed innovators looking to contribute to the future of their country’s defense. Complicated rules and slow processes mean that there is a mismatch between the timelines of government-run projects and those of fast-moving, commercial organizations. Meanwhile, the lack of data standards within and across government agencies remains a problem. It’s often exceedingly difficult for contracted companies to locate, access, and leverage pertinent data. Working with the government poses great challenges to many private companies, and the relationship between public and private sectors in the U.S. is difficult to develop and maintain. Private organizations are often disinclined or unable to attempt to bridge this gap.
This needs to change—and sooner rather than later. The sparing AI bets placed by some investors are now paying off, but the future of our country’s technological investments still seems uncertain.
Federal AI research and development programs should not be viewed as zero-sum investments to be measured against other priorities. The country will need to deal with the budgetary implications of the $6 trillion we’ve spent to combat the pandemic. But exponential technologies are not a mere luxury in good times; they can provide critical efficiencies, global competitiveness, and cost-effective national security in the hardest of times.
If governmental AI investments can move beyond research prototypes to real, functional programs through a smarter acquisition strategy, these promising technologies could be used to alleviate many of our most difficult problems—even in the face of overall budget reductions. Spending smarter doesn’t imply spending more overall.
DoD’s Joint Artificial Intelligence Center organization can play a critical role in all this. There is a need for aggressive advocacy within the Pentagon to identify problems that can be paired with practical AI solutions in line with DoD’s priorities of joint warfighting requirements, readiness, efficiency, and cost control. Investments in the JAIC should be increased and protected.
Civilian agencies face similar challenges with AI. Though less platform-centric than DoD, agencies like the Department of Energy or Health and Human Services must also hew to fiscal realities and become far more efficient in how they manage and employ their personnel or prepare for future pandemics.
Even though scale and efficiency don’t usually go hand in hand, given the budgetary pressures it is incumbent on agencies to reduce costs without harming performance. This is where AI has shown great promise in the commercial world. Natural language processing analyzes large amounts of text automatically and has made commercial call centers tremendously more efficient. The same systems could help cope with the unprecedented citizen demand for government assistance. Such capabilities are crucial to the next generation of AI-enabled public health initiatives, including global pandemic surveillance and emergency economic assistance that will reach federal, state, and local agencies.
To succeed, the U.S. government must redouble its efforts to cultivate real partnerships with private technology companies that understand both AI and governmental needs, and can provide an undiluted, full-spectrum focus. Granting key AI contracts to longtime prime contractors or consulting shops may be necessary, but it isn’t sufficient. Acquisition professionals must go beyond their comfort zones to seek out expert AI partners committed to an across-the-board public sector transformation. To that end, AI companies are heavily focusing on government customers and the best ways to serve them.
This is a critical moment for the country—not only in its response to this pandemic, but for our future with AI. Navigating these tough times requires not just survival in the present, but a credible vision for a brighter future and a desire to engage with more action and less talk.
Robert O. Work is Chairman of the Board for SparkCognition Government Systems, and sits on the Board of Directors of Raytheon Technologies Corporation and several smaller defense-related firms. He is a Senior Fellow at the Johns Hopkins University Applied Physics Laboratory.
Lisa Disbrow sits on the Board of Directors for SGS, and on the Board of Directors of Mercury Systems, Perspecta, and BlackBerry, and several other defense-related companies. She is a Senior Fellow at the Johns Hopkins University Applied Physics Laboratory.