September 13, 2018 at 07:52PM
Go to the source
To ensure continuity across the 17 disparate agencies that make up the IC, program managers looking for AI solutions are adhering to the concept of AIM, or Augmenting Intelligence using Machines.
“With that, we’re auditing routine functions that analysts, curators, collectors have manually done in the past to free them up to perform more intelligence,” La’Naia Jones, IC deputy chief information officer, said at the annual Tech Trends Conference hosted by the Professional Services Council. “Common sense, right? Sounds much easier than it is.”
Jones noted information gathering is part of the mission sets of each IC component, meaning lots of data.
“There is no big data analytics without a large amount of data. We need the machines to help us read and understand the data and we need the humans to look at it overall,” she said. “The amount of data is just increasing exponentially.”
During a roundtable with reporters, Jones explained the nature of AIM, which a way of thinking, not a technology or program office. But that thought process is essential to the success of any AI initiative, she said.
“When you speak of AIM, it’s a concept, it’s a vision, it’s an idea, it’s a way of doing business,” she said. “We’re partnering within the architectural division to leverage the reference architectures so that … whatever they set up is compliant and shareable amongst the other agencies … It’s not a stovepipe solution.”
Many agencies and programs are also working with academia, particularly at the local level, Jones said.
“In everything we do, we’re always looking at: how can we collaborate, how can we integrate, how can we work together, how can we team?” she said. “How can we better work this problem or these issues together so that way we don’t create the same problems?”
During her presentation, Jones offered three examples of how AI is being used within the AIM framework by IC agencies today.
At the National Security Agency, AI is being used to augment human language processing. Jones said commercial tools have only come so far, and NSA has unique problems, including complex languages and degraded audio.
“NSA has retooled the process to integrate AI across the board,” she said, changing “the way the workflow is performed and to create a continuous training set for the machines.”
The National Geospatial-Intelligence Agency is using AI tools to sort large amounts of imagery and pull out specific targets, such as enemy safehouses and airfields—work that has been done manually by analysts in the past. The new tools enable those analysts to do more of what they’re best at: analyzing data.
“We’re moving the analysts from counting things to thinking about it,” Jones said. “Of course, a computer can tell us an airfield is empty but it can’t tell us why.”
Jones also offered an example from the National Reconnaissance Office, which is using AI to find the kernels of useful information in the troves of data the agency collects daily.
“This is helping to detect activity and automating tasks for expansive collection systems,” she said. “We can assure that most of the data that we collect is in the best interest of our analysts, to provide the best intelligence to the decision-makers.”
As with the other IC agencies, NRO’s efforts fall in line with the AIM process, Jones said.
“It’s looking at what your mission goals are—what you’re trying to achieve—and then can you use automated intelligence, can you use AIM—this philosophy, this process, this way of doing business—in order to improve that? Can any efficiencies be gained? Can we use that in order to push the envelope further?” Jones said.