Skip to main content
U.S. flag

An official website of the United States government

Remarks as delivered by NGA Director Vice Adm. Frank ?Trey? Whitworth for DoD Intelligence Information System (DoDIIS) Worldwide Conference

Remarks as delivered by
Vice Adm. Frank “Trey” Whitworth, USN
Director, National Geospatial-Intelligence Agency
DoD Intelligence Information System (DoDIIS) Worldwide Conference
“Chaos to Clarity: Leveraging Emerging Technologies”
December 12, 2023
Oregon Convention Center
Portland, Oregon

Hey, team. Thanks a lot, Sarah [Maj Sarah Answine, USAF, Defense Intelligence Agency (DIA) CIO Senior Representative to U.S. Strategic Command].

And I gotta make fun of last year for just a second, because I was so inspired to hear the walk-up music. Last year if you were here, the walk-up music was Yo-Yo Ma. And so when I was asked: “Hey, do you want some walk-up music?”, I’m like: “Absolutely. Put the Foo Fighters up there.” And I think it’s appropriate to have Dave Grohl singing “Times Like These.” If you know that particular song, and you know those lyrics, I think it’s appropriate. Also, Dave Grohl is from Springfield [Virginia], where we have our headquarters – one of our main buildings, of course. And he also has an affinity for the Pacific Northwest. And so this is really an important opportunity for NGA, and also for all of us to be together.

What a tough act to follow. Thank you to both the PDDNI [Principal Deputy Director of National Intelligence Dr. Stacey Dixon] as well as [DIA Director] General [Scott] Berrier for just enlightening us.

It’s an honor to be here on behalf of nearly 15,000 hard-chargers of the combat support agency known as the National Geospatial-Intelligence Agency, or NGA for short – that wild light, blinding bright.

Even though we’re balancing targeting, warning, and safety of navigation on a 24/7 basis – and even though we’re constantly building our warfighters’ and policymakers’ understanding of the pacing challenge of China, I&W [indications and warning] and force disposition affecting Russia’s unprovoked invasion of Ukraine, and steady accounts of events unfolding in the Levant – and even though we’re already busy putting geospatial context into deliverables, we brainstormed as a flat NGA team in the SCIF [sensitive compartmented information facility] this morning, during our daily 0545 Pacific time O&I [operations and intelligence brief]; and yes, I was there – there is nothing more important right now than communicating with the thousands of you here, and ultimately virtually, at this DoDIIS Worldwide Conference.

Let’s get right down to it. Today, I’m going to focus on three aspects of NGA’s warfighting tradecraft: First, how we orchestrate the collection of GEOINT for the combatant commands and National Command Authority. Second, how we integrate artificial intelligence, machine learning, and data dissemination for the benefit of those warfighters – capabilities that resonate with our conference theme today of moving from “Chaos to Clarity.” And third, how we embrace innovation at NGA, and more broadly in the National and Allied Systems for GEOINT, with notable mission impact.


INFORMED COLLECTION ORCHESTRATION
So, let’s start with informed collection orchestration.

Now, if you’re not familiar with “informed collection orchestration,” join the club! The Defense Intelligence Enterprise and IC [Intelligence Community] writ large innocently went for decades not really knowing that NGA professionals task overhead GEOINT collection 24/7. And let’s be honest, NGA never actually sought credit in this regard. It was, and still is, a natural process, kinda like breathing. It just happens, with little fanfare.

Our GEOINTers align their working knowledge of observable behaviors and historical precedents with multiple variables, like the requirements of combatant commanders – with whom we’re located, deployed theater and operational commanders – also with whom we’re located, weather forecasts, orbits, maintenance impacts, and of course national priorities.

Now, if you think national priorities trump COCOM [combatant command] priorities in GEOINT collection, think again. That’s a myth. We’ve crunched the numbers, and the overlap of those priorities is pervasive. We know both the numerator and the denominator of collection requirements for every COCOM; and while I cannot relay them here, please trust me when I tell you they’re impressive.

If you think NGA prioritizes analytical needs over operational ones, think again. That’s also a myth. Products do not outcompete our warfighters. Theories do not outcompete targets. We are a combat support agency for a reason.

If you think NGA orchestrates sensing without waiting for time-late instructions – that’s true! When you’re balancing collection opportunities against China, Russia, Hamas, North Korea, Iran – and I could go on – there’s no time to dither.

Honestly, much of our informed collection management happens while commanders are asleep. We know their intent, and we move out. What’s wonderful about this decentralized execution is this: It works, as those commanders fall in on our collection decisions and exploitation output; and I receive no complaints. None. And I checked again this morning at 5:45.

So, what’s the secret ingredient here? Well, let’s go back to the nomenclature I’ve adjusted by adding the word “informed.” Informed collection orchestration means our role isn’t limited to statute, policy, or practice. It’s rooted in our long-stare for context, our target expertise, our predictive instincts, and our mastery of GEOINT exploitation. To be clear, informed collection orchestration is a global responsibility, and it’s becoming more complex by the day.

Given our location in every COCOM and their headquarters, NGA is the organization that maintains a comprehensive understanding of GEOINT collection risks across all combatant commands. So accordingly, we’re also the agency that can dynamically exercise risk mitigation across all of them as well.

This becomes increasingly important as more data arrives. Newer sources of commercial imagery, like synthetic-aperture radar, or SAR, coupled with allied approaches to collection, give us more options and permutations for collection to prioritize than ever before.

All this requires us to provide warning, not only to U.S. decision makers and mission partners, but to our allies around the world. And we need to be able to provide near real-time data – whenever and wherever it’s needed – for immediate impact and decision advantage.

And by the way, we can also squeeze in support to natural disasters: Oil spills, typhoons, earthquakes, even wildfires and floods at home, if and when our support is requested by other federal agencies.

With all of this, we have to check every day to make sure our readership, our consumers – policymakers, warfighters, first responders – we have to make sure they’re getting what they need.

My last point here is this: We don’t have the luxury of only working crises. We have global responsibilities that morph with new realities. So while we’re working hard to support our Ukrainian partners, giving them the max availability of commercial imagery and analysis, and managing the threat to U.S. forces in the Middle East; we also still need to maximize understanding of our pacing challenge: China. That includes understanding the state of their force posture, their R&D, and their global activities.

Fortunately, we multitask pretty well, because we have the GEOINT experts who turn pixels and data into insightful analysis for our global customers – from the combatant commands of course, all the way to the White House. And those experts do it every day, without missing a beat.

That’s our big picture, our mission: Providing “clarity” amidst the “chaos.”

 

AI / MAVEN / ML / CV
Now, let’s turn to what I’m sure is on many of your minds today: Artificial intelligence and machine learning.

There’s a book that came out earlier this year called Knowing What We Know: The Transmission of Knowledge: From Ancient Wisdom to Modern Magic. It describes how before written language came along, oral storytellers were responsible for all the wisdom that was passed down. The book basically argues that we no longer “think” the same way our ancestors did.

And this is a timely point, given that things are about to change even more drastically. Because when it comes to AI – ChatGPT in particular, and its eventual evolution into something even more advanced – writing by machines is happening right before our eyes.

There are some experts who predict that AI will be more revolutionary than the invention of the printing press, the Industrial Revolution, the aerospace industry, computers, and the internet combined.

While that might bring some social adjustments, we should also expect progress for immense good. Examples: Crops will be more disease-resistant, water desalination will improve, medicine will improve.

Deep Learning, with AI models patterned on the human brain, and "neural networks" of nodes interconnected in many layers, is on the horizon. At NGA, we embrace this opportunity, and we’re getting ready.

Last year, I spoke at this conference about how data volumes in GEOINT increased so much that they created a new challenge that I termed “deluge control.” In the GEOINT world, we simply have too much data to sift through, the old-fashioned way at least, and that’s why we need AI.

In the last year, we’ve taken over operational control of most of what was known as Project Maven. It was DoD’s AI pathfinder, originally started in the Office of the Under Secretary of Defense for Intelligence and Security. Our portion, the GEOINT lines of effort, was renamed NGA Maven. And I’m pleased to tell you it became a program of record last month. So, we’re privileged to be a leader in this portion of the development and application of AI, for both the IC and DoD. And on a more practical level, it’s something we have to do, because we’re responsible for analyzing a staggering amount of data. And we’ll detail that in a few minutes.

Fortunately, Maven’s scan is impressive. It can quickly fuse enormous amounts of data from across disparate data sets, teeing up meaningful opportunities. We’ve worked hard with the combatant commands to integrate AI into their workflows, accelerating operations and decision speeds. We’ve also increased the fidelity of targets, improved geolocation accuracy, and refined our test-and-evaluation process. Plus, we’ve ensured Maven models can run on other machine learning platforms.

But there will always be concerns, which is why we need AI assurance and governance processes that promote secure and ethical AI development, and instill confidence to meet mission.

Six weeks ago, President Biden signed Executive Order 14110 on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence. It underscores the extraordinary potential of AI to solve urgent challenges, while balancing against the possibilities of harm – like fraud, bias, disinformation, and risks to national security.

We realize that we at NGA need to be at the forefront of ensuring the ethical and responsible use of AI. So we’re leaning forward. We simply can’t afford not to, since our enemies aren’t pausing. But we’re also doing it all carefully: Establishing a certification program that we call the GEOINT Responsible AI Training program, or GREAT for short.

Let me give you a few unclassified examples of ways that AI is now being used at NGA, and will most likely be used to support our mission in the near future.

Our current AI primarily consists of machine learning and computer vision, or CV for short, both of which we’ve been doing for years, to help our analysts keep up with deluge control. We use ML and CV in a variety of ways: From something as simple as tipping and queuing our analysts, to look at imagery, of course – to AI-powered global mapping – to AI-enabled social media monitoring of terrorist organizations and known adversaries, using location markers to help verify videos and images posted to social media.

We’ve also worked with the National Ice Center, to create automated charting of changes in Arctic sea ice, used for both navigation and to track and monitor climate change. 
We’re sharing lessons of our AI, ML and CV projects, and we’re building algorithms for NGA Maven that will be usable across DoD and the entire IC.

We’re also using AI to increase the speed and capability of humanitarian efforts, as we support DoD and other federal agencies. This is especially important when extreme weather events – like heatwaves, drought, typhoons, and flooding – increase the demand for disaster relief responses around the world, as well as here at home.

We also play a key role in anticipating and mitigating climate change, and its resulting security challenges. For example, we support “climate informed intelligence” as routine parts of the decision-making process. That includes the location and status of climate vulnerable countries’ food and water resources and population migration patterns.

And here’s a quick illustration of why we need machine learning and computer vision, to sort through all the chaos: You’re probably familiar with facial recognition and the tech that accompanies your going through a customs or some sort of a check before you’re about to get onto an international aircraft. For those algorithms, a face can be about 80 percent of the image that’s being analyzed – 80 percent of what we call the field of view.

What we do is so much harder. By comparison, many of the objects we’re trying to develop and identify, and the behaviors that we’re trying to identify, represent two one-hundred-thousandths of a percent. So again, comparing 80 percent of a field of view to two one-hundred-thousandths of a percent of an image.

Last week, I did a podcast and I mentioned a different number, and it’s simply because there are so many zeroes that we lose track. We double-checked. Don’t reverse-engineer this. We’re also offering a little bit of fuzz here, so that we don’t get into whether we’re being completely and utterly specific on the resolution of overhead imagery, but take my word for it: This is infinitesimal. 

And if that’s not hard enough, what we’re often looking for is a mobile adversary that’s trying to hide.

So, we really need to develop relevant innovative operational concepts, deploy cutting-edge capabilities, and invest in capabilities we’ll need well into the future.

And let’s not forget about that pacing challenge of China, and what the Chinese are doing with AI. In 2016, the Chinese People’s Liberation Army established the Strategic Support Force to centralize their space, cyber, and electronic warfare missions and capabilities under a single command. And China’s military is using AI to collect, fuse, and transmit massive quantities of data to speed their processing, and create their information advantage over the U.S. and our allies. They also enjoy a very direct relationship with their own commercial industry, incorporating these assets into their architecture.

So even as we respond to crises of the moment, from Europe to the Middle East, we will always stay focused on China.

Speaking of industry: For us, commercial computer vision detections have helped us automatically identify improvised airstrips. And they’re feeding automated strategies for more dynamic collection. Commercial analytic services are also monitoring oil tanks, and providing new insights into economic stability. We recognize that a holistic effort isn’t something we can do alone, so we rely a great deal on partners from academia, industry, and allies. And that’s been a theme this morning.

And lastly on AI, I would ask you to visit our NGA booth – it’s booth #921 – and Dr. Heather Martin will actually be discussing our “Way of AI” tomorrow at 12:30, and Thursday at 9:30.

 

DATA DISSEMINATION

Moving over to data, another theme this morning.

Last year, I spoke a bit about how we get our data to those who need it at the tactical edge. As all of you know, having the best data in the world is useless, if you can’t securely move it, where and when it needs to go. And visual data is so dense that it’s far more challenging to disseminate than other kinds.

NGA ingests, moves, analyzes, and stores more data than any other agency, so it’s important to do it quickly, and in the way our consumers need it. Our communication pipes are world-class in scale and throughput; and they’re so vital to warfighting and crisis management, that we must preserve them in a contested environment. And that’s where the concept we’ve developed for the future of GEOINT access and delivery comes into play: The Joint Regional Edge Node, or JREN.

JREN widens the delivery pipe for dissemination at the theater-level, working alongside our two established systems: Odyssey at the component-level, and GEOINT Grab and Go, or G3, at the tactical-level. We’re very close to initial operational capability, which will increase resiliency, reduce latency, and help us exceed user requirements.

Over the past few years, we’ve actually increased our infrastructure bandwidth between core sites by an order of magnitude of ten. Our current main system, Odyssey, has been what’s called “one-to-many” infrastructure. You can imagine a wheel hub, with spokes connected to that central hub. Well, JREN and Odyssey together are moving towards a “mesh architecture” that provides a “many-to-many” capability. It’s basically several hubs connected to each other, and also to the many spokes sitting on those peripherals.

If you’re interested in a deeper dive on this, my teammate Chuck Bellinger will be at the booth on our floor, and he can give you additional insights in two sessions, tomorrow and Thursday.

 

RECENT ITEMS OF INTEREST

Transitioning out of data, let’s finish with part three, and some new technical innovations we’re working on at NGA.

First is our new cybersecurity strategy. The goal with this is to deliver a standardized and secure GEOINT enterprise through improved risk management, zero-trust principles, and advanced defense. It employs data and AI for security decision-making, providing continuous monitoring and threat identification.

Second, there’s an upcoming field test by our Warfighter Support section later this month and in January, near the Washington D.C. area. They’re going to simulate a tactical unit that’s conducting land-based collection, processing, exploitation, and dissemination – completely disconnected from their tactical operations center – using a radio-based mesh network to disseminate information. It’s based on real situations that our U.S. Special Operations Command partners have faced. And it’s meant to empower a user in the field.

That same group, our Warfighter Support section, is also currently in Ecuador of all places. A volcanic dam there eroded and collapsed, which triggered massive erosion and sediment flows and damaged critical infrastructure, and now threaten a hydroelectric plant that provides a third of Ecuador’s power. So our team is using their small UAS [unmanned aircraft systems] drone kits to map this catastrophic geohazard on a river. In addition to the Ecuadorians, they’re also working with Oak Ridge National Laboratory and the U.S. Army Corps of Engineers.

We also have what we call our Frostbyte team, and that’s “byte” with a Y. And they just returned doing field work in the Arctic, studying permafrost melt. They’re research scientists who are partnering with DoD and others in Alaska, to develop automated data-driven methods to monitor permafrost-related terrain instability from space. Their goal is to enable faster and more accurate assessments of infrastructure challenges in the polar regions. It’s about mitigating risks to human health, flora and fauna, roads, pipes, and other infrastructure sitting atop melting permafrost. And of course, it’s also about national security, because the Arctic is of huge strategic interest these days.

We have another initiative designed to anticipate risk that will start early in the new year, and it’s called the Global Fishing Forecast Challenge. It’s a contest to engage creative commercial thinkers to identify innovative ways to use unclassified data to forecast illegal, unreported and unregulated fishing activity, known in the business as IUU fishing. We’re partnering with the National Security Innovation Network on this. Of note, a multi-organizational judging team of maritime experts will select ten companies or teams to pitch their recommended solutions, and they each win at least $25,000. A $500,000 first prize will be awarded in late spring. I believe finding new data sources and methods to forecast the movement of what are considered “dark” vessels would be a big plus there.

It’s also a pleasure to announce that NGA was recently awarded the IC’s first Frontier Project, with the DoD High Performance Computing, or HPC, Modernization Program. The team will use HPC resources to develop the largest and most advanced AI model for Earth observation-based querying. It’s a four-year effort that will involve millions of hours of computing time, along with dedicated systems administration support, and will advance the state-of-the-art of AI.

And last but not least, I’d like to make just a bit of news today, by announcing that we’re planning to release a new request for proposals, or RFP, next month. It’s to purchase new commercial GEOINT data services from multiple vendors. The resulting contracts are called LUNO A – Luno Alpha – and they’re intended to help us meet the increasing demands for commercial GEOINT. The goal is to enable NGA to do two things: Acquire commercial GEOINT object detections, and leverage industry analytics and automation, in areas of national security interest, of course. I wish I could tell you the amount, but let’s just say it’s significant.

 

CLOSING

All right, this brings me to the end of my prepared remarks. I think I’ve got just a couple of [minutes] to plug just several of my wonderful colleagues and friends.

First, our NGA CIO [chief information officer] Mark Chatelain is on the CIO panel, tomorrow at 11. Our GEOINT Services director Jim Long will be in Breakout 7 tomorrow at 1:30 for the “Cloud Services” panel, and again at 4:30 for the “Rapidly Emerging Tech” panel.  And on Thursday morning at 8, one of our cybersecurity engineers, Evan Kehayias, will be on the “Zero Trust” panel in Breakout 7, where he’ll talk about NGA's data as a strategic asset, and the need to defend it as such.

Of course, I also invite you all to stop by our NGA booth, again 921, to see some great demos – all designed to support our mission and industry partners.

I can’t tell you how privileged I am, to get to work with such dedicated intelligence professionals. Their value is demonstrated like clockwork every day, 24/7; but especially for me, it’s such a privilege to listen to them at 5:45 a.m. Pacific, with our Ops and Intel brief. It’s flat and it’s fast, in the same way I’m accustomed to teams as we started our days at JSOC [Joint Special Operations Command], USCENTCOM [U.S. Central Command], USAFRICOM [U.S. Africa Command], and I could go on. And that’s the same way we began today.

Even with two enormous regional crises occurring simultaneously, NGA and the entire GEOINT community have risen to meet the needs of our allies, friends, and partners.

This conference is all about turning chaos into clarity. What we do at NGA is call balls and strikes. We’re dedicated to providing clarity. We convey that essential distinction between fair and foul, safe and out, friend and foe, combatant and noncombatant, where and when – that our leaders and warfighters need.

And we’ll do everything we can to be that vanguard of distinction for both DoD and the IC – in “Times Like These,” and time again.

Thank you.