How the CIA is using generative AI — now and into the future

From human intelligence collection to information gathered in the open, the CIA is leveraging generative artificial intelligence for a wide swath of its intelligence-gathering mission set today, and plans to continue to expand upon that into the future, according to the agency’s AI lead.

The CIA has been using AI for things like content triage and “things in the human language technology space — translation, transcription — all the types of processing that need to happen in order to help our analysts go through that data very quickly” as far back as 2012, when the agency hired its first data scientists, Lakshmi Raman, the CIA’s director of AI, said during an on-stage keynote interview at the Amazon Web Services Summit on Wednesday in Washington, D.C.

On top of that, AI — particularly generative AI in recent years — has been an important tool for the CIA’s mission to triage open-source intelligence collection, Raman said.

“Imagine all of the news stories that come in every minute of every day from around the world. That is also the data that we are neck deep in, and that we’ve got to leverage AI in order to help us to triage,” she said, adding that generative AI has been successful in helping the CIA “classify and triage open-source events to help us search and discover and do levels of natural language querying with that data.”

Raman said the CIA, despite its clandestine reputation, was not insulated from the “generative AI zeitgeist” that has captivated the world since late 2022. That has resulted in other use cases like “search and discovery assistance, writing assistance, ideation aids, helping us with brainstorming, helping us generate counter-arguments.”

In an interview with FedScoop on the Daily Scoop Podcast earlier this year, Raman described how the CIA navigates using commercial generative AI tools due to the classified, sensitive nature of its work, saying that the agency wants to “use those solutions whenever we can.” However, there are cases with classified data that the agency’s AI specialists must build what Raman called “exquisite solutions for our more bespoke problems.”

With this adoption of generative AI comes added work around creating an environment that it can thrive in and a new, evolved approach to risk, she said Wednesday.

“In addition to the actual pointy edge-of-the-spear tools, right, we also need the foundational capabilities to help us run and deploy the models,” Raman said at the summit. “We have catalogs for our models, as well as our training data sets. So it’s really about us creating that entire ecosystem for us to be able to leverage this to advance our mission and national security.”

In thinking about risk, Raman said one of her main focuses has been: “How will our users be able to use these technologies in a safe, secure and trustworthy manner?”

“That’s about making sure that they’re able to look at the output and validate it for accuracy,” she said. “And being bias-free, right? We work very closely with our general counsel, as well as our Office of Privacy and Civil Liberties, to think about those types of activities. We want to make sure that our users feel very comfortable with it, so they understand that what they’re seeing is AI-generated output.”

What’s ahead for the CIA and generative AI

Raman has a number of things on her mind as she ponders the future of generative AI within the CIA, from how it could become a normal part of each of the agency’s five directorates to the way the country’s top adversaries are also building AI capability for their advantage. But ultimately, she doesn’t see it changing the CIA’s mission beyond it serving as “another tool in our toolbox to help us drive the human-machine teaming that is now so necessary for us to do our jobs.”

“We don’t see AI or technology as anything that is going to replace our workforce. But we see it as something that is going to heavily augment our workforce, and help us accelerate our velocity in completing tasks so that we can leverage our minds for higher-order work,” she said. “So we definitely see AI as something — and generative AI particularly — as something that can be incorporated into workflows all across the agency to help us augment our own judgment and move through our workflows faster.”

As such, Raman said she anticipates AI and AI-focused work roles to permeate functions across the CIA.

“We have a very broad mission at the CIA. We have five directorates that do everything spanning from creating objective analysis, to doing collection, to creating very unique, exquisite technical tools, to spearheading our focus on the digital landscape, but also to doing administrative and logistical functions,” she said. “So that’s a lot of opportunities for people at the agency, in the AI space particularly.”

While Raman didn’t describe her agency’s development of AI capabilities and an AI workforce as a race against adversaries like China and Russia, she did describe how those nations’ pursuit of developing AI capacity for their political, economic and military advantage ultimately impacts the CIA on its journey.

“There’s a lot for us to be thinking about with respect to, you know, how we are reacting to our adversaries’ plans, use and intentions with respect to emerging technology, more generally, and AI very specifically,” Raman said.