‘The US can’t lose’: Pentagon plows ahead on AI despite warnings

Politico | Leo Shane III

The Pentagon is quickly bringing artificial intelligence into all of its military areas in order to make sure that adversaries such as China and Russia do not achieve a significant technological edge. Troy Meink, who is the Secretary of the Air Force, stated that the United States cannot afford to lose this high-stakes race and has rejected the suggestions made by critics to put the development on hold.

As members of Congress rush to add regulatory safeguards to the forthcoming annual defense policy bill, the strategic risks involved are huge. Critics such as Senator Richard Blumenthal have warned that rogue autonomous agents could damage sensitive networks, but defence officials contend that the most dangerous course of action would be to avoid adopting AI. At the same time, Europe's allies back the speed of this move because their stockpiles of munitions have been depleted by the war in Iran and this has increased the demand for high-tech military alternatives. In the end, the Pentagon's Chief Digital and AI Officer, Cameron Stanley, stresses that the military will continue to be an AI-first force while keeping human judgment at the heart of all autonomous decisions.

Comment

The Pentagon's quick deployment of algorithmic warfare tools marks a basic departure from the usual process of validating capabilities through the Joint Capabilities Integration and Development System towards faster operational prototyping. This change is shown by the Replicator initiative, which avoids the traditional procurement routes in order to put into service thousands of autonomous attritable systems. The US military is now accepting considerable cyber vulnerabilities by focusing on speed of deployment rather than carrying out thorough software verification, in order to keep up with the rapid modernisation of the Chinese military.

The reason behind this change in doctrine is that the Chief Digital and Artificial Intelligence Office is deploying continuous software integration pipelines directly to the tactical edge networks. Instead of having to wait for hardware acquisition programmes that take many years, the US Indo-Pacific Command is now receiving iterative updates to its algorithms in order to counter specific capabilities in the People's Liberation Army's electronic warfare efforts. This approach of continuous deployment greatly changes the way the US Joint Force checks autonomous target-recognition algorithms before they are used in combat in the Western Pacific.

Strategic Question for Discussion

Would the capability of the US Joint Force to guarantee algorithmic reliability in contested situations fail if the Replicator initiative keeps on avoiding the usual Joint Capabilities Integration and Development System validation, or is it possible for continuous software pipelines to alleviate these risks?

The fact is that avoiding the existing validation procedures will certainly lead to software vulnerabilities, but since continuous integration pipelines enable quick patches whereas static procurement cannot, this gives an advantage. I believe that the practical value of being able to rapidly deploy large numbers of units in the Western Pacific will in the end outweigh the worries about algorithmic unpredictability. As a result, the US Joint Force will probably tolerate higher initial failure rates in order to achieve rapid tactical adaptation against the People's Liberation Army's countermeasures.

Share your assessment in the comments below.

To read in details click here