Artificial Intelligence in the Intelligence Community
April 30, 2020 in Technology & Tools
By Justin Fuhrmann
Every month we explore the various tools and technologies that impact individuals and organizations. This month we are taking a closer look at Artificial Intelligence (A.I.), which will have a profound impact on the future of the U.S. Intelligence Community (IC). I was fortunate enough to attend the most recent Kalaris Intelligence Conference (“the conference”) and hear from leading A.I. and IC experts on just what the future may look like and what we all can do to prepare and shape this technology. There is not a unanimous consensus of what the largest challenges are today or what lies ahead, but there were several opinions where common themes provide guidance of what we must consider for the future.
Why Use A.I. At All?
We must consider why the IC needs to utilize A.I. at all. A litany of science fiction movies and television shows have demonstrated that there are numerous concerns when it comes to A.I. What are the ethical concerns, how do we avoid A.I. absorbing existing human biases, and how do we ensure that we don’t create a real-life version of Skynet? One of the best reasons why the IC needs to utilize A.I. is out of necessity. Stephanie O’Sullivan, Former Principal Deputy Director of National Intelligence, speaking at the conference stated that, “Whichever country comes to dominate A.I. will own the future.”
The National Geospatial-Intelligence Agency (NGA) has already encountered the necessity to incorporate A.I. into its work. Government and commercial satellites are now so numerous that every day over 1.5 million images are taken of the Earth’s surface. NGA’s job is to take those images and analyze the differences between them day after day. However, analyzing this level of data would require around 6 million Geo-Analysts in order to properly inspect all the images. As Former NGA Director Robert Cardillo explained to CBS’s 60 Minutes in 2018, “While that number [of Analysts] may be exciting to the Director of NGA, it is less exciting to the American taxpayer.”
To solve this concern of so much data coming in each day, A.I. is needed. A.I. is required to analyze these images each day and then to flag for a human analyst when an imagine needs a closer look to determine if the change from day to day is the result of regular human activity (e.g. commercial) or if it represents a threat or concerning development. This is how A.I. enables NGA to fulfill its vital mission daily. To adapt to this new reality NGA has shifted to recruiting data scientists while simultaneously recruiting individuals capable of thinking about bringing human analysis to data collection.
“We as a society are drowning in information without being able to discern it’s importance, relevance and truthfulness.” Sue Gordon, Former Principal Deputy Director of National Intelligence, shared that sentiment in her keynote address at the 2019 Kalaris Intelligence Conference. Global IT systems are more open than ever before and are vulnerable to manipulation. In an era of returning to great power competition, rival totalitarian governments seek to undermine public trust in the U.S.’s systems and institutions. If we are currently drowning in data, A.I. can lend us a figurative life raft.
The Path Forward
We already see examples of totalitarian governments employing A.I. to violate human rights. Ms. Gordon suggests that to address this concern and ensure similar violations do not occur in the U.S. we must harvest data with civil liberties and privacy built in. Technology alone cannot meet this need; human talent must be involved as well. Ms. Gordon suggests the need to establish an ethical framework for A.I. in government because it is vital for ensuring society trusts that the IC is acting in a trustworthy manner.
As for the IC and how it can rise to meet this emerging challenge of working with A.I., Ms. Gordon suggests that the IC needs to undergo a mindset shift from, “What do we need to know” to, “how do I assess what I see.” In order to assess what we are seeing in a rapidly evolving world, she predicts that more data will be a necessity, and that data must be made useable with A.I.’s help. Humans alone cannot undertake this endeavor because, just as NGA receives too many images every day to have human eyes look at all of them, decision makers throughout the world are confronted with more data than they can process and make sense of on their own.
Mike Horowitz, professor of political science and the interim director of Perry World House at the University of Pennsylvania says that, “A.I. will help to ensure the continued U.S. dominance in national defense,” and many of the speakers at the conference shared that sentiment. Regardless of the challenges to get there, A.I. is being developed by other nations and the United States must develop its own capabilities if we are to play a role in how this technology will progress . Many of the speakers at the conference talked about the need to ensure that strong protections and security are built in A.I. from the start. The internet lacked this type of focus on security when it was initially developed and we pay the price for this routinely.
Beyond national defense and the IC, how else might A.I. impact our lives? A.I. may play a significant role in enhancing human education. It may help to bridge the gap between how machines and people understand each other. It may allow for developing augmented reality to help us be more productive in our work and our personal lives.
Some speakers at the conference shared that they are more concerned about A.I. becoming big brother rather than the terminator. This apprehension is, in part, driven by a concern that A.I. will be prone to acquiring the biases of its developers. There are ways to make A.I. more accurate and robust and to not take on racist and sexist biases and this topic is also explored in one of our blogs from earlier this year. Developers need to think critically about their inputs so A.I. does not make bad decisions based on prejudiced inputs. Having diverse developers matters to ensure prejudices are identified before moving to implementation.
Utilizing developers with diverse backgrounds is an area where Mr. Horowitz believes the U.S. is well positioned because the U.S. can attract the best talent and combine it with the benefits of diversity, “We have a political philosophy which embraces this reality. We are a nation of allies and partners which helps us advance technologies. We have the best research universities on the planet. I would much rather have the U.S.’s problems versus anyone else. We tend to focus on our deficiencies rather than our very real built-in advantages.”
Parts of the IC is already utilizing A.I. to fulfill their mission and more agencies are likely to follow in the future as national leaders seek solutions to make sense of overwhelming data. By taking a lead in developing A.I. the U.S. has a greater opportunity to influence and shape the development of this vital technology. What do you see as the challenges and advantages with A.I.? Are you convinced A.I. can be used as a force for good, are we bound to build HAL 9000 or will the result be somewhere in-between?
Has your organization or industry been impacted by artificial intelligence? Share your thoughts about A.I. with us on LinkedIn!