Before I Enter the Room: Using AI to Prepare for an Intelligence Conference
Next week, I am attending a major intelligence and national security conference as a journalist.
The program is dense. Over a day and a half, senior government officials, military leaders, intelligence professionals, researchers and private-sector technology companies will discuss everything from artificial intelligence and quantum security to the Arctic, facial recognition, missile defense, autonomous warfare and lessons from Ukraine.
There is an exhibit hall. There are plenaries. There are competing breakout sessions. There are 15-minute lightning rounds and networking breaks. There is a reception.
And there is no possible way for one journalist to see everything.
So before deciding where I should be, I did something that has increasingly become part of the way I think through complicated subjects.
I asked AI.
But I didn't simply paste the schedule into ChatGPT and ask it which sessions sounded interesting.
I asked a more interesting question.
Based on our previous conversations, my research interests and the subjects I have been following, where should I be?
The Difference Between Search and Context
I have spent a lot of time talking with AI.
Some of those conversations have been mundane. Others have involved subjects I have returned to repeatedly: artificial intelligence, China and Taiwan, Russia, space, national security, military exercises, public displays of state power, surveillance, nuclear energy and the changing relationship between technology and government.
More recently, I have been examining the return of large-scale public spectacle and visible state power, comparing developments in the United States with military exercises and displays in Russia, China and Europe.
I have also spent considerable time thinking and writing about AI itself, including the increasingly complicated question of what constitutes human authorship when AI becomes part of the intellectual and editorial process.
That history mattered when I handed over the conference schedule.
Instead of asking AI to tell me which speakers were most famous or which sessions sounded most important, I asked it to evaluate the program in relation to the questions I am already asking.
The result was remarkably specific.
Building a Journalist's Priority List
The first problem was triage.
On Wednesday afternoon, for example, two breakout sessions occur simultaneously.
One examines operational readiness and resilience in the Arctic. The other is titled “Quantum-Ready: Navigating the 2027 Mandate for National Security.”
Both interest me.
AI recommended quantum.
The reasoning wasn't simply that quantum computing is important. The panel brings together representatives from the White House Office of the National Cyber Director, the Cybersecurity and Infrastructure Security Agency, the Center for a New American Security, the University of Maryland and SandboxAQ.
The more useful journalistic questions became apparent: Why 2027? What actually needs to happen by then? Who is prepared? Who isn't?
The recommendation wasn't a conclusion about what my story should be. It identified where questions worth asking might be concentrated.
That distinction matters.
Golden Dome, Facial Recognition and the Future of Intelligence
Other priorities emerged quickly.
A 15-minute presentation by Clearview AI on facial recognition became a must-see, not necessarily because a corporate presentation should be accepted at face value, but because it presents an opportunity to examine the expanding role of private companies in supplying capabilities once associated primarily with governments.
How does the company talk about privacy?
Accuracy?
Oversight?
Scale?
Permissible use?
Immediately afterward comes a fireside chat with Lt. Gen. Brian Gibson, vice director of Golden Dome for America.
That became another priority.
Golden Dome sits at the intersection of several subjects I have been following: missile defense, space, sensors, artificial intelligence, data integration, military procurement and the Trump administration's larger defense agenda.
AI also suggested something obvious that can easily disappear when staring at a conference agenda:
Try to talk to him afterward.
The panel matters.
The hallway might matter more.
Turning a Program Into a Reporting Schedule
We eventually converted the conference program into something closer to an assignment plan.
Wednesday begins with reconnaissance of the exhibit hall, followed by the opening program and keynote luncheon.
At 1:30 p.m., I go to the quantum panel.
The afternoon includes time on the exhibit floor, followed by Clearview AI at 3 p.m., Golden Dome at 3:15 and the Service Intelligence Priorities plenary at 4:45.
That last panel puts intelligence leadership from the Army, Navy, Air Force and Coast Guard on the same stage.
AI suggested that instead of simply taking conventional notes, I compare them.
What does each service say about China? Russia? Space? Artificial intelligence? Autonomy? Data? The private sector? Workforce?
If several independently emphasize the same issue, the repetition itself becomes information.
And then, at 6 p.m., comes something AI ranked above almost everything else:
The reception.
Not because receptions are inherently important, but because journalism is still a human business.
A conference is not only what happens onstage. It is what people say afterward, what they mention over coffee, what they are willing to discuss when they aren't holding a microphone and what leads to the next conversation weeks or months later.
Thursday: When the Themes Begin to Converge
Thursday morning begins with a Strategic Priorities Panel featuring senior leadership from the Defense Intelligence Agency, National Reconnaissance Office and National Security Agency.
That is followed by a lightning-round presentation with a title that immediately caught our attention:
“The Ethical Imperative to Use AI: Preserving Tradecraft Across the Intelligence Community.”
The wording represents a potentially significant rhetorical shift.
The question is no longer simply whether using artificial intelligence is ethical.
The proposition embedded in the title is that there may be circumstances in which not using AI is the less responsible choice.
That is worth listening to very carefully.
Then comes another choice between simultaneous breakout sessions.
One examines the evolution of government information sharing.
The other is called “Mass, Autonomy, and the Ukraine Model.”
AI put the second one at the top of my list.
Ukraine has become a laboratory for questions that extend far beyond the war itself: inexpensive drones, autonomous systems, commercial technology, distributed sensors, rapid iteration and what all of that means for militaries historically organized around extraordinarily expensive platforms.
The question I am taking into that room is deceptively simple:
What does “mass” mean now?
Does military mass mean soldiers, ships and aircraft?
Or could it increasingly mean thousands of relatively inexpensive autonomous systems?
I don't know the answer.
That's why I'm going.
Using AI Before the Reporting Begins
Perhaps the most useful part of the exercise wasn't scheduling at all.
It was identifying patterns I might otherwise have encountered separately.
Quantum security. Facial recognition. Golden Dome. Artificial intelligence. Autonomous systems. Commercial technology. Intelligence sharing. Private defense companies.
Viewed individually, they are conference sessions.
Viewed together, they begin to suggest larger questions about what the American intelligence and national security apparatus is becoming.
That doesn't mean I should arrive with my story already written.
Quite the opposite.
One of the dangers of preparing too aggressively is confirmation bias. If I enter the conference determined to prove a thesis, I may miss the more interesting thing happening directly in front of me.
So another part of my reporting plan is deliberately open-ended.
I am going to keep track of repetition.
What words keep appearing?
What problems are mentioned by completely different organizations?
What technologies appear repeatedly in presentations and exhibit booths?
What is everyone suddenly selling?
What are government officials asking industry to provide?
And what subjects seem conspicuously absent?
I also plan to photograph the exhibit hall itself: signage, slogans, demonstrations, screens and the visual language companies use to sell the future of intelligence.
Institutions communicate through images as well as words.
Sometimes the walls tell you what the panel doesn't.
Four Stories to Watch For
Before attending a single session, AI helped me identify four possible stories.
They aren't predetermined articles. They are hypotheses to carry into the reporting.
1. The Intelligence Community Has Crossed the AI Threshold
The question may no longer be whether America's intelligence agencies should adopt artificial intelligence.
The more interesting question may be whether AI is becoming so embedded in intelligence work that refusing to use it is increasingly understood as an operational liability.
The “ethical imperative” presentation gives me a place to begin testing that idea.
2. From Golden Dome to Drone Swarms: America's New Architecture of Defense
Golden Dome, service intelligence priorities and the Ukraine autonomy discussion may reveal different pieces of the same transformation.
Sensors, satellites, artificial intelligence, autonomous systems and distributed networks increasingly connect domains that were once discussed separately.
The question is whether the conference participants themselves describe them as parts of a larger architecture.
3. The Privatization of Intelligence Capability
Clearview AI. Maltego. SandboxAQ. Accrete. TekSynap. Broadcom. BAE Systems. Saronic Technologies.
The conference program contains government officials alongside an enormous private technological ecosystem.
Walking the exhibit floor may help answer a provocative question:
What intelligence capabilities that once required a government agency can now be purchased from a company?
4. What America's Intelligence Establishment Is Worried About in 2026
This may ultimately be the most interesting story because I cannot know the answer before I arrive.
Instead, I can listen.
If the DIA talks repeatedly about one problem, the NSA mentions it, military intelligence leaders return to it, researchers discuss it and vendors are building products to solve it, the pattern itself becomes worth examining.
The conference may tell me what the story is.
The Schedule
For now, my reporting plan looks like this:
Wednesday, August 26
10:00 a.m. — Exhibit hall reconnaissance
11:00 a.m. — Welcome remarks
11:30 a.m. — Opening keynote luncheon
1:30 p.m. — Quantum-Ready: Navigating the 2027 Mandate for National Security
2:30 p.m. — Exhibit hall and networking
3:00 p.m. — Clearview AI facial recognition lightning round
3:15 p.m. — Golden Dome fireside chat
4:15 p.m. — Hallway reporting and networking
4:45 p.m. — Service Intelligence Priorities
6:00 p.m. — Reception and source building
Thursday, August 27
8:00 a.m. — Breakfast and networking
9:00 a.m. — Strategic Priorities: DIA, NRO and NSA
10:30 a.m. — Networking and speaker follow-up
10:40 a.m. — The Ethical Imperative to Use AI
11:00 a.m. — Solving the AI Data Bottleneck
11:15 a.m. — Mass, Autonomy, and the Ukraine Model
Possible Articles
The Intelligence Community Has Crossed the AI Threshold
How artificial intelligence is moving from an optional technological capability toward an expected component of intelligence tradecraft.
From Golden Dome to Drone Swarms: America's New Architecture of Defense
How missile defense, space-based sensors, AI, autonomy and lessons from Ukraine may be converging into a different model of American defense.
The Privatization of Intelligence Capability
A look at the expanding ecosystem of companies selling technologies that increasingly shape how intelligence is collected, analyzed and acted upon.
What America's Intelligence Establishment Is Worried About in 2026
A reported examination of the threats, technologies and vulnerabilities that repeatedly surface across agencies, military services, researchers and industry.
The important thing is that none of these articles exists yet.
AI helped me read a schedule. It helped me connect that schedule to months of previous conversations and research. It helped me prioritize my limited time, identify people I may want to speak with, organize a reporting strategy and recognize several possible stories.
But next week, I have to walk into the rooms.
I have to listen.
I have to ask the questions.
And, perhaps most importantly, I have to be willing to discover that the real story is something neither of us anticipated.
Process disclosure: The author used AI for brainstorming, research support, organization, drafting, editing, and formatting and layout while producing this journalism. Responsibility for the ideas was shared; responsibility for the final form was entirely human. The resulting scores were AI Thought 59/100, AI Mechanical 63/100, and Human Originality 40/100. Human approved.