AI not Allies: The US Army prioritises future C2

The last 30 years has seen a proliferation of command-and-control systems in military headquarters. Despite promises to the contrary, there has been no correlation between the arrival of such systems and the size of staff required by commanders; arguments for reducing staff numbers because of new systems rarely meets with reality or the ever-growing demands on HQs for differing mission requirements, reporting chain demands, and widening responsibilities loaded onto HQ personnel.

However, the aim of C2 systems is not to reduce headcount; rather, the right systems allow for greater situational awareness of the battlefield (or campaign) and – hopefully – better command decisions.

Delivering almost near real-time SA is not a difficult concept – an essential element of The Quest for Certainty (Martin Van Creveld) – and so it might be surprising that so many commercial offerings cannot deliver. A few select companies can really parse and interface data from multiple systems, fewer still have the User-Interface that works for military personnel.

An issue for several militaries has been their focus on national level solutions for these C2 systems at the cost of interoperability with allies and partners; it is usually these same institutions that talk a lot about the weight they place on enhanced civil-military co-operation but do not consider scalability options for C2 systems with those agencies when it is required (often at extremely short notice).

Scalability and interoperability are key facets of C2 systems in an era of great power competition. In most combat operations, the need to integrate ad-hoc forces, formations and units, as well as external partners will be core to fighting and resilience response. Yet few of the commercial offerings have an ability to work across multiple security classifications and domains, with a UI that feels intuitive and a system that is reliable and dependable.

The functionality and efficacy of HQ C2 systems is highly dependent on the code that underpins the software as much as the data that it exploits. Code written by operations-orientated designers has a very different ‘feel’ and utility to it than core code developed primarily for, say, intelligence work. The priority, bias and emphasis placed on certain elements of information and data will be treated in different ways and with different outcomes depending on how the system itself was originated. The ’better-decisions’ that C2 systems are employed for in HQs thus depends – to a disproportionate extent – on the origins of the system and the background of the coders. In simple terms, if the C2 system was designed for intelligence, logistics, or engineering, it will not be as effective at C2 decision support as one designed and developed by operators for this distinct purpose.

The US Army seems to be in exactly this position. In Project Maven, the US Army seems to be hoping that an intelligence backbone can be developed into a boutique system that will meet the needs of commanders and staff in an expanding remit that seems to have started incorporating C2 functions. Whilst developers may promise much in terms of scalability and interoperability, there is less confidence amongst partners and allies that this can be realised – certainly in the short term. It is also less clear that Maven solutions can survive in heavily contested EM environments, or even in poor natural conditions: there are reports that snow and heavy foliage can hinder its ability to accurately identify targets. If intel systems cannot perform in these conditions, the vulnerability of it as a C2 system should be subject to much greater scrutiny.

There is no doubt that the US Department of Defense feels it is a race to militarise AI with the Chinese PLA. This, it seems, is behind the drive for Maven at the price of everything else. As the US Army has been forced into de-emphasising a demand for interoperability with partners in order to prioritise AI (with all the unrealised promise which that might bring). Now the choice has been made, commanders can only hope that Maven realises its goals within the next 18 months (in keeping with the US view on potential conflict timelines with China). And that they don’t need to integrate or work too closely with allies to deliver mission success.

All of this doesn’t bode well for the ‘fight tonight’ mission unless commanders can demand a change to the priorities of the DoD and Congress.