U.S. Transportation Command sees potential in the way commercial artificial intelligence solutions can help it leverage vast amounts of data and improve its logistics operations, but is still in the early stages of deciding how it will be best used, the command’s commander said Oct. 2.
“We’re trying to be disciplined in the approach” to AI, said Army Gen. Steve Lyons during a Wednesday Defense Writers Group breakfast in Washington, D.C.
“Because much of our logistics data is in the unclassified space, there’s opportunities really to leverage commercial technology in this area,” he noted. To him, the high-end goal of harnessing AI capabilities would be to network together disparate systems to better utilize their data and enable better decision-making forecasting and maintenance reliability, among other efforts.
While the issue of which platforms to incorporate is best left up to the individual armed services, Lyons and his team are focused on how to integrate such commercial solutions at the command-and-control (C2) level, and have developed an “enterprise data environment” as a proof of principle, he said.
Lyons’ predecessor as USTRANSCOM commander, now-retired Air Force Gen. Darren McDew, spearheaded a “sprint to the cloud initiative” where the command moved a limited number of systems into the cloud, Lyons said. Work continues to move the entire architecture over to the cloud, and that is expected to take some time, he added.
Shifting products to the cloud “is not a magic snap your fingers and all of a sudden you’re there; it’s a very arduous journey over time,” Lyons noted. “What we did is we took about a half-dozen use cases, and we’re trying to work through the use cases to say, ‘How can we really leverage the data that we have in meaningful kinds of ways? And then, eventually, you start to hit a tipping point where you start to move and gain momentum, and that’s essentially how we’re approaching it.”
One challenge he has seen involves the need to “cleanse” the data to make it useable across the new cloud environment, he said. The command itself controls nearly 100 information technology systems and even more exist across the broader joint deployment enterprise.
“They have all been evolved and developed in their own kind of ways for the right kind of reasons. But now as you look at it from a top perspective of how you want to enable strategic decision-making [and] resource allocation, there’s a lot of work to be done to bring these into … a cloud-computing native environment where we can leverage the data and get to a higher opportunity for machine-learning,” Lyons said.