1005 Alcyon Dr Bellmawr NJ 08031
+1 484-446-2566
support@bmcpros.com
Every day, hundreds of chat messages between pilots, crew members and controllers of Air Mobile Command’s 618 Air Operations Center (AOC). These controllers indicate a fleet of one thousand wide aircraft, juggling variables to determine the route to fly, how much time it will take to refuel or load, or can fly these tasks. Their mission plan enables the U.S. Air Force to quickly meet global national security needs.
“For example, it takes a lot of work to get missile defense systems around the world, and this coordination used to be done by phone and email. Now we are using chat, which creates opportunities for AI to enhance our workflow,” said Joseph Monaco, strategy director for No. 618, AOC.
618 AOC sponsors Lincoln Labs to develop these artificial intelligence tools through a project called “Conversational AI” Technology (CAITT).
Meeting with laboratory researchers to discuss CAITT during a visit to Lincoln Laboratory at Scott Air Force Base, Illinois, Colonel Monaco, Colonel Tim Heaton and Captain Laura Quitiquit. CAITT is part of a broader effort to transform artificial intelligence technology into a major Air Force modernization program called Next Generation Information Technology Improvement (NITMRE).
The type of AI used in this project is natural language processing (NLP), which allows models to read and process human language. “We are using NLP to map key trends in chat conversations, retrieve and reference specific information, and identify and contextualize key decision points,” said Courtland Vandam, a researcher at the AI Technology and Systems Group at Lincoln Labs. CAITT includes a suite of tools that leverage NLP.
One of the topic summary is one of the most mature tools that extracts trending topics from chat messages and is in a user-friendly display to highlight key conversations and emerging issues. For example, a popular topic might read: “Crews lack Congo visas, potential delays.” This entry shows the number of chats related to the topic, summarizes the key points of the conversation in the bullets and links back to a specific chat exchange.
“Our missions are very time-dependent, so we have to quickly synthesize a lot of information. This feature does prompt us to focus on the effort,” Monaco said.
Another tool for production is semantic search. This tool improves the search engine of the chat service, and the tool will currently return an empty result if the chat message does not contain every word in the query. With new tools, users can ask questions in natural language formats, such as why a specific aircraft is delayed and smart results are obtained. “It incorporates a neural network-based search model that understands the user intention of the query and goes beyond term matching,” Vandam said.
Other tools in development are designed to automatically add users to chat conversations related to their expertise, predict the basic time it takes to unload a specific type of cargo from an aircraft, and summarize key processes in regulatory documents as a guide for operators to plan their missions.
The CAITT project evolved from the DAF-MIT AI accelerator, with a three-fluid effort between the MIT, Lincoln Laboratory and the Department of Air Force (DAF) to develop and transition AI algorithms and systems to promote DAF and society. “Through the NITMRE project, by participating in the AI accelerator, we realized that we could do something innovative with all the unstructured chat information in the 618th AOC.”
As laboratory researchers promote the prototype of their CAITT tool, they have begun converting it into the Department of Defense software provider 402ND SOFTWOLE ENGINEERING GROUP. This group will implement the tool into the operating software environment used by the 618th AOC.