Defense Secretary Pete Hegseth this week released strategies meant to focus the Pentagon’s “alphabet soup” of innovation organizations and proliferate artificial intelligence—moves that experts say could provide the structure needed to make the military’s efforts to integrate and field new technology more effective.
Issued in two separate memos, the strategies seek to “transform” the Defense Department’s approach to developing new technology and operational concepts. The AI directive aims to funnel AI models and tools into Pentagon processes, planning, and operations.
“Becoming an ‘AI-First’ warfighting force requires more than integrating AI into existing workflows,” the memo states. “It requires reimagining how existing workflows, processes, [tactics, techniques, and procedures], and operational concepts would be designed if current AI technology existed when they were created—and then reinventing them accordingly.”
It also calls for expanding the department’s AI compute power, making data more accessible across the department, and puts the Chief Digital and AI Office in charge of “eliminating blockers”—any policies or other barriers that get in the way of experimenting with and fielding AI-enabled systems.
Hegseth’s restructuring of the Pentagon’s innovation organizations, meanwhile, is an attempt to bring order to an ecosystem that many DOD officials and outside experts say has grown unwieldy in recent years. Mike Brown, a former director of the Defense Innovation Unit, said that while organizations like DIU, the CDAO, and the Strategic Capabilities Office were created with good intentions, they’ve lacked a clear focus.
“We kind of had 1,000 flowers bloom, and it became confusing,” Brown told Air & Space Forces Magazine. “I think this is bringing them together into some rational design.”
The document makes clear that authority for setting technical direction for the department lies with Undersecretary of Defense for Research and Engineering Emil Michael, now the department’s sole CTO, who will oversee the Pentagon’s six innovation organizations—the SCO, DIU, CDAO, the Defense Advanced Research Projects Agency, the Test Resource Management Center, and the Office of Strategic Capital.
It also defines three lines of effort as focal points for three of the organizations:
- DIU will focus on commercial capability integration
- DARPA will focus on military-unique technology development
- SCO will focus on operational concept innovation
Further, it calls on the military services to craft plans within 90 days for how their labs and innovation hubs will organize around these three goals.
“The intent and policy are simple: Unify the innovation ecosystem led by a single CTO tasked to modernize the department and align innovation organizations around outcomes that matter for the warfighter,” the memo states.
Avoiding Novel AI Programs
As with all strategy documents, the impact of Hegseth’s orders lies in its implementation, and defense experts said they’ll be watching closely to see if the Pentagon can achieve and maintain the discipline the memos dictate.
Central to the AI and data strategy is the idea that AI shouldn’t be treated as a separate product, but as a part of many of the workflows, processes, and systems the military relies on. Bryan Clark, director of the Hudson Institute’s Center for Defense Concepts and Technology, said the Pentagon has a tendency to create standalone programs for novel technologies and centralized offices, like the CDAO, to oversee them.
“This approach turns that around and says, this is like electricity—it’s a fundamental technology that we’re going to need in a lot of different systems,” Clark told Air & Space Forces Magazine.
As the department puts its new strategy to work, Clark said he’ll be watching to see if the CDAO and the services can avoid falling back on old habits.
“Part of what’s happening here is trying to make the CDAO much more of an organization focused on data, because that’s something they need somebody to be trying to harness, as opposed to being the organization that’s the central clearinghouse for all AI programs,” he said. “I’ll be interested to see if the department’s able to stay disciplined and avoid creating AI programs and instead focus on creating programs that use AI.”
The strategy also doesn’t discuss an issue the military has long faced when trying to build more autonomy into its systems—trust between a machine and its operator. Clark said he’s interested to see how the military validates the accuracy and judgment of the AI capabilities it adopts.
“The foundational issue is: How confident am I that its results are useful and accurate,” he said. “Do I have to have some secondary effort to test that, evaluate it, before I can employ the system for decision making?”
Metrics for Innovation Organizations
Hegseth’s plan to restructure the Pentagon’s innovation offices follows a sweeping set of acquisition reforms issued in November that dissolved the Pentagon’s arduous requirements process, replaced stovepiped program offices with portfolio managers, and called on the services to adopt commercial-like buying practices.
Those reforms, according to Brown, set a strong foundation for the deliverables Hegseth now wants to see from the department’s innovation offices.
“I’m very enthused about the combination, when you put all of this together,” he said.
One key metric is whether this more focused, defined approach can quickly yield results, Brown said.
“I’m sure across the innovation ecosystem, each of these organizations will have its own metrics,” he said. “We’ll want to see if they can execute.”
That’s true not just for defense-wide organizations like DIU and the SCO. Brown said it’s clear from Hegseth’s mandate that service labs and innovation hubs will also need to prove they can produce tangible results.
“The goal for everyone should be: How do we get more scale and better technology and capability to the warfighters,” he said. “If I’m running a lab, I should be continuously looking for, where’s the commercial market for providing this. That, I don’t think, was as strong a direction previously.”
