The agency’s long tech wish list includes capabilities that can predict human movement and new computing architectures to crunch its mountain of data.
The intelligence community is preparing for a busy research year with up to a dozen new projects, many of which lean on artificial intelligence and a special tailor-made quantum offering for the National Security Agency.
Catherine Marsh, the director of the Intelligence Advanced Research Projects Activity, previewed a range of the agency’s research offerings that are expected to take shape this year during a keynote address at the Intelligence National Security Alliance’s Spring Symposium April 11.
Marsh said the goal is to kick off 10 to 12 new projects this year, including one that attempts to link “digital fingerprints” to a text’s author. In the coming weeks, Marsh said the agency will release a broad agency announcement for HAYSTAC, a program that seeks to “figure out human movement” using AI-reasoning engines that can identify abnormal patterns while generating normal ones.
The IARPA director’s comments come after the Office of the Director of National Intelligence released its four-year tech investment plan, which ranked artificial intelligence among the most sought after capabilities, including expanded data, cyber, computing and sensor capabilities.
One project, called WRIVA, seeks to make the intelligence community’s imagery surveillance more like it’s portrayed in Hollywood, using "Walkthrough Renderings for Images of Varying Altitudes" to improve surveillance detection routes and allow users to go “anywhere in the world without going out into the wild,” Marsh said.
But they don’t have the data.
“Where our agents and capabilities need to go, they're not the exotic places like Paris, where there's plenty of data and imagery, right? We're talking about places, which are probably not the nicest places in the world and how we create that [in] near real time so that we're able to send people into remote areas without really going there?” Marsh said.
The RIVA project will aim to use software algorithm-based systems to develop photo-realistic and navigable site models using imagery, data, and “small amounts” of metadata, she said.
On the computing front, IARPA expects to announce award winners for its Advanced Graphical Intelligence, Logical Computing Environment, or AGILE, program in late April or early May, Marsh said. The program sought to explore entirely new high-performance, high-efficiency computing architectures that can handle the ever-growing amount of data the IC ingests.
“All that data that we talked about, we have to think about doing compute differently. Absolutely differently,” Marsh said.
And when it comes to quantum computing, Marsh said the agency is still working on the budding technology.
“So quantum is still very early and nascent. And we've been involved with quantum since the beginning and stand up of IARPA. [A program] that we have ongoing right now is the LOGIC program, which—this is how early and nascent the science is—we want to demonstrate one logical qubit. … It hasn't been demonstrated yet. And so being able to do that and show that that can then translate into that capability is critically important,” Marsh said.
Marsh also said that there would be a new quantum focused-program coming out later this year called ELQ, which stands for extensible logical qubit. The effort is currently in development and geared toward providing next generation capabilities for NSA.
This article was updated April 14 to clarify the acronym for IARPA's "Walkthrough Renderings for Images of Varying Altitudes" program.