Every day, hundreds of chat messages flow between pilots, crew, and controllers of the Air Mobility Command’s 618th Air Operations Center (AOC). These controllers direct a thousand-wide fleet of aircraft, juggling variables to determine which routes to fly, how much time fueling or loading supplies will take, or who can fly those missions. Their mission planning allows the U.S. Air Force to quickly respond to national security needs around the globe.
“It takes a lot of work to get a missile defense system across the world, for example, and this coordination used to be done through phone and email. Now, we are using chat, which creates opportunities for artificial intelligence to enhance our workflows,” says Colonel Joseph Monaco, the director of strategy at the 618th AOC, which is the Department of Defense’s largest air operations center.
The 618th AOC is sponsoring Lincoln Laboratory to develop these artificial intelligence tools, through a project called Conversational AI Technology for Transition (CAITT).
During a visit to Lincoln Laboratory from the 618th AOC’s headquarters at Scott Air Force Base in Illinois, Colonel Monaco, Lieutenant Colonel Tim Heaton, and Captain Laura Quitiquit met with laboratory researchers to discuss CAITT. CAITT is a part of a broader effort to transition AI technology into a major Air Force modernization initiative, called the Next Generation Information Technology for Mobility Readiness Enhancement (NITMRE).
The type of AI being used in this project is natural language processing (NLP), which allows models to read and process human language. “We are utilizing NLP to map major trends in chat conversations, retrieve and cite specific information, and identify and contextualize critical decision points,” says Courtland VanDam, a researcher in Lincoln Laboratory’s AI Technology and Systems Group, which is leading the project. CAITT encompasses a suite of tools leveraging NLP.
One of the most mature tools, topic summarization, extracts trending topics from chat messages and formats those topics in a user-friendly display highlighting critical conversations and emerging issues. For example, a trending topic might read, “Crew members missing Congo visas, potential for delay.” The entry shows the number of chats related to the topic and summarizes in bullet points the main points of conversations, linking back to specific chat exchanges.
“Our missions are very time-dependent, so we have to synthesize a lot of information quickly. This feature can really cue us as to where our efforts should be focused,” says Monaco.
Another tool in production is semantic search. This tool improves upon the chat service’s search engine, which currently returns empty results if chat messages do not contain every word in the query. Using the new tool, users can ask questions in a natural language format, such as why a specific aircraft is delayed, and receive intelligent results. “It incorporates a search model based on neural networks that can understand the user intent of the query and go beyond term matching,” says VanDam.
Other tools under development aim to automatically add users to chat conversations deemed relevant to their expertise, predict the amount of ground time needed to unload specific types of cargo from aircraft, and summarize key processes from regulatory documents as a guide to operators as they develop mission plans.
The CAITT project grew out of the DAF–MIT AI Accelerator, a three-pronged effort between MIT, Lincoln Laboratory, and the Department of the Air Force (DAF) to develop and transition AI algorithms and systems to advance both the DAF and society. “Through our involvement in the AI Accelerator via the NITMRE project, we realized we could do something innovative with all of the unstructured chat information in the 618th AOC,” says Heaton.
As laboratory researchers advance their prototypes of CAITT tools, they have begun to transition them to the 402nd Software Engineering Group, a software provider for the Department of Defense. That group will implement the tools into the operational software environment in use by the 618th AOC.
Source: https://news.mit.edu/
