New Air Force Vice Chief of Staff Gen. Jim Slief has made leveraging data one of his priorities and is seeking industry support to improve how the service collects and uses data.
Speaking to a crowd Thursday at an AFCEA luncheon, Slife outlined what he sees as the Air Force's biggest hurdles in leveraging data for operations, training, maintenance and more. In general, he said, most of the data the Air Force collects is not tagged or analyzed, and many of the Air Force's databases are isolated from each other, making it difficult to use that information to create accurate service-wide solutions. He said it cannot be used.
“I know that a lot of the solutions to these things exist as point solutions. We need a more holistic approach to this,” Slife said. “We can’t just have ‘Vendor A’ and ‘Product A’ solve ‘Problem A.’ We need to think about data and AI challenges holistically across departments.”
A big problem within the Air Force is that much of the information collected “never sees the light of day” for several reasons, chief among them being the sheer volume of raw data that cannot be manually transmitted and analyzed. Slife said this is because it is difficult to do so. Additionally, strict classification policies can prevent who can access the data and where it can be used.
Slife cited the F-35 stealth fighter as an example. The aircraft collects terabytes of data from radars, sensors, electronic warfare systems, and communications equipment for each mission. But instead of having a way to collect the information and automatically classify it for future use, he said, the data is often deleted or overwritten.
“That data incorporates lessons learned. There was a wingman who did the wrong thing, there was a bad radio call, there was a signal we had never seen before,” Slife said. “We need it to incorporate it into our future missions and train our algorithms with the truth we need for accurate AI models.”
Because the Air Force often lacks network capacity to transfer large amounts of data, the service copies information to physical hard drives and transfers it elsewhere. Still, the data must be manually indexed, a process that takes several weeks, he added.
“When we run out of time, we often dump a lot of that data somewhere in the cloud, a so-called 'data lake,' but that's unindexed data,” he explained. “These data lakes actually contain more unusable data than usable data. Therefore, these lakes become 'data swamps.' ”
He said the Air Force is trying to process the vast amounts of data coming from new platforms while also finding ways to access information from legacy systems. Most of the service's aircraft use the military standard 1553 data bus, which manages and transmits all onboard information on the platform. Slife said the data could be used to rebuild missions, identify failed parts or detect cyberattacks, none of which is recorded.
Another problem, he said, is that the key data used by aircraft maintainers is spread across multiple databases owned by different organizations that don't communicate with each other. These stovepipes prevent the Air Force from performing accurate predictive analysis that can be used to retain conventional aircraft longer.
“We now have all the data we need to accurately predict aircraft component failures. But it's fragmented and the databases don't talk to each other, so efficiency is reduced accordingly. ” he said. “Hospice care is the most expensive form of medical care, and many of our fleets provide hospice care.”
Slife said he plans to work with the Air Force Chief Data and Artificial Intelligence Office to address data inefficiencies in the service.
He also issued a “call to action” for the industry.
“All the data we need to remain the best Air Force in the world is at our fingertips,” he said. “I'm not asking for a teleportation machine. I'm not asking for a magnetic flux capacitor. I'm not asking for anything beyond the realm of possibility. These are all things that our It is within reach, but we need innovative solutions to capture, catalog and classify it.”