In the next 90 days, the U.S. Indo-Pacific Command, in collaboration with corporate partners and the Pentagon, will evaluate how generative AI tools akin to ChatGPT could assist commanders in making faster battlefield decisions, particularly against advanced adversaries like China.
Radha Plumb, the head of the Defense Department’s Chief Digital and Artificial Intelligence Office (CDAO), stated the objective is to conduct tests within the INDOPACOM area of responsibility using specific Navy scenarios. This prototype will involve partnerships with companies like Anduril and Palantir to enhance decision-making capabilities by reducing response time and expanding situational awareness for commanders.
A senior industry official involved in the CDAO initiative described the project as more than just a new product. It aims to establish a framework among a consortium of next-generation tech companies and military commanders. This will facilitate a better understanding of how generative AI can be integrated into operations and may involve either building new solutions or improving existing tools.
The CDAO’s exploration of generative AI represents a new focus on utilizing advanced AI technology to expedite commanders’ comprehension of battlefield dynamics and command issuance during major conflicts. This initiative builds upon prior pilot projects aimed at applying large language models in areas such as military healthcare, logistics, and contract management.
Outgoing White House National Security Advisor Jake Sullivan emphasized the necessity to integrate AI capabilities across all defense operations, advocating for rapid advancements not only in weapons systems but also in logistics and supply operations. He pointed out that the upcoming administration should prioritize this integration.
One significant challenge in implementing generative AI in military operations is the substantial computing power required, as highlighted in the recent National Defense Authorization Act, which mandates an expansion of high-performance computing infrastructure.
Plumb noted that traditional government computing methods have not sufficed for advanced AI applications, typically relying more on unclassified computing rather than classified environments. Generative AI demands considerable computational resources to function effectively, particularly for tasks requiring the integration of diverse data sources.
Concerns persist, however, regarding the potential consolidation of power among AI companies and how that could affect the Pentagon’s leverage. Sullivan acknowledged the collaborative efforts of defense tech firms as a promising development as long as the government maintains control over procurement and budgeting.
Regarding the new defense consortium, Sullivan expressed that as long as oversight remains strict, these collaborations could foster positive change in the defense contractor landscape. Plumb detailed the CDAO’s Open Data and Applications Government-owned Interoperable Repositories program (DAGIR) as a means to allow the Pentagon flexibility in selecting products and companies without necessitating complete redesigns.
Despite these advancements, a senior Defense Department official cautioned about the challenges the government faces in maintaining leverage in a highly consolidated digital market. If not managed correctly, this consolidation could lead to defense contractors prioritizing profit over the actual needs of warfighters, potentially depriving them of essential tools and technologies.