‘All hands on deck’ moment in the US-China AI race: Pentagon’s former digital chief

The U.S. may cede its technological superiority to China if it does not speed up the development and deployment of artificial intelligence, the former head of the Pentagon’s office for accelerating AI adoption said Monday. 

During an emerging-technology summit at the Johns Hopkins University Bloomberg Center, Radha Plumb — who served as the Defense Department’s chief digital and AI officer during the final year of the Biden administration — said competition with China, in particular, necessitates “an all-hands-on-deck moment” for the U.S. government and industry partners, given Beijing’s military and economic strength. 

Whether it’s firms building cloud environments or physical infrastructure, Plumb said “all of these pieces need to work together in a coherent ecosystem” to provide a competitive edge. But part of the problem, she added, is that the federal government is not fast enough when it comes to purchasing and deploying new technologies and “we don’t have good models for scaling it.”

She warned that, when it comes to building out advanced AI capabilities in particular, “I think we do have a moment here, and I worry sometimes that it’s passing us by.”

Plumb used the rollout of 5G capabilities to underscore her point, noting that the deployment of the advanced cellular network also depends on procuring the necessary hardware and applications to create an integrated AI ecosystem.

“If that is ceded to Chinese companies and/or Chinese government solutions, that will mean we’ll use 5G Huawei networks on Huawei devices using DeepSeek applications with our data plus the Chinese data, and that’s not an ecosystem that lends itself to U.S. — or even U.S. allies’ and partners’ — advantage,” she said.

Congress and federal officials have already moved to cut off Chinese companies like Huawei from doing business in the U.S., citing national security concerns. Lawmakers directed the Federal Communications Commission in 2020 to establish a “rip and replace” program to reimburse communications providers for the removal and replacement of devices made by Huawei and ZTE Corporation, although allocated funding has not matched the on-the-ground needs.

Novel and expedited uses of AI tools in military conflicts, however, have also underscored the importance of speeding along the testing and deployment of U.S.-made capabilities. 

Pointing to Ukraine, which uses low-cost weaponry, such as explosive-packed drones, to counter Moscow’s invasion—Plumb also said that simply upping investments in new capabilities will not drive innovation as much as moving to use and deploy the actual technologies. 

“I think the lessons are, basically, you need a much faster iterative cycle that takes operational need to technology, to testing, ideally as close to real testing as you can get, and then back around again,” she said, adding that “the faster you can do that cycle with real testing of real technology in the hands of real warfighters, the faster you get that innovation and creation cycle to drive technology improvement.”

Working to advance broader uses of AI tools in the U.S. through carefully crafted guidance, however, also remains a key part of this discussion. Plumb said that “a bifurcated approach” to federal policies around AI could also harm the deployment of these capabilities.

She said this includes efforts to implement policies where AI is “either entirely unregulated and unfettered and unchecked in how it uses data, what it uses data for and what applications it can use; or it is extremely heavily regulated, such that the use of the technology goes underground or doesn’t get adopted in kind of a very pragmatic way.”