Technology companies integrating AI-enhanced back-end military functions with defence companies’ missiles will be the first step in a new partnership between the two sectors. Neil Thompson reports.
Western efforts to integrate artificial intelligence (AI) into military operations and technology development (including drone and missile targeting) are no longer in their early stages. Projects for implementing AI into the armed forces like the US military’s Project Maven, which began in 2017 and is expected to report later this year, are beginning to bear fruit.
Though AI breakthroughs for missile targets are still around four years away, according to Courtney Manning, director of AI Imperative 2030 at the American Security Project, defence companies are making a number of innovations using AI that can be applied to upgrade already-existing weapons technologies like missile systems.
These innovations will accelerate in the coming decades as Western militaries face renewed challenges from geopolitical rivals like Russia and China.
“With the recent downsizing of the federal workforce and increased participation of tech moguls like Elon Musk in the current administration, I now expect that [US] technology firms themselves will have a greater impact on the way AI is incorporated in US [weapons] systems than any [government-lead] Maven-like initiative,” Manning said.
“The US may be slipping behind in new AI patents filed and firms breaking into the market, but [it] remains the global leader in battlefield technology.”
Tech firms’ growing involvement with missile defence
By far the most significant impact of AI for Western militaries at present will be in improvements to the ‘back-office’ functions of the military enterprise, rather than modifications to missile weapons.
These functions include summarising documents, finance, human resources, and spotting patterns in collected intelligence, translation, and signals and images. This will improve the detection systems that governments rely upon to decide on first strikes or activating missile defences.
In particular, AI image recognition from complex scenes recorded via video from systems like satellites will steadily improve. Specialised tasks like change detection of images, combining image and other data to find patterns of behaviour, and other intelligence functions will also get better as AI systems are integrated into legacy military structures.
These AI integrations will require the deep participation and engagement of the US AI industry, particularly major companies like Microsoft, Google, and OpenAI, and Western defence contractors.
At present, though China is catching up, the US firms hold the keys to the most advanced AI models on the market. The US Department of Defense will work with industry on joint development projects and programmes, with significant oversight from leading engineers in the private sector.
Defence contractors will also need to be involved to ensure that AI integration makes sense with existing hardware, like intercontinental ballistic missiles (ICBMs) and missile defence systems.
For example, in December US defence firm Anduril announced it is working with AI developer OpenAI on missiles and drones that can better use the mass volumes of data collected by defence satellites and installations.
“Anduril builds defence solutions that meet urgent operational needs for the US and allied militaries,” said Brian Schimpf, co-founder & CEO of Anduril Industries. “Our partnership with OpenAI will allow us to utilize their world-class expertise in artificial intelligence to address urgent air defence capability gaps across the world.
“Together, we are committed to developing responsible solutions that enable military and intelligence operators to make faster, more accurate decisions in high-pressure situations.”
First strike capacities, AI & safety concerns
The US is currently likely going to be cautious about introducing AI directly into its offensive systems, according to Dr Iain Boyd, director of the University of Colorado’s Center for National Security Initiatives. Instead, it will continue to rely on a large arsenal of conventional nuclear-capable ICBMs as the cornerstone of its nuclear deterrence strategy.
Dr Tom Stefanick, a non-resident senior fellow at the Brookings Institution’s Strobe Talbott Center on Security, Strategy, and Technology, concurs that currently a nuclear strike will probably not wipe out all enemy missile systems in a major power like the US, China or Russia, so global deterrence remains intact, at least for now.
“A reliable and rapid first-strike capability with thermonuclear weapons is unlikely to result in the complete elimination of an opponent’s nuclear arsenal [a present],” he determined. “I think a reliable first-strike capability that results in near-elimination of a devastating retaliatory capability is not feasible”.
Meanwhile, Manning forecasts that AI-enhanced missile targeting is still estimated to be about four years away, meaning until then offensive missile systems are unlikely to be any more accurate in a first strike attack scenario than they are at present.
Instead, AI is more likely to have an impact in the management of how countries deploy the variety of weapons that they now have access to, to overwhelm and confuse their enemies’ defences, deterring them from believing they could protect themselves in the event they attacked first. Countermeasures to sensing in the physical battlespace from the seabed to outer space are very active areas of research, said Stefanick.
A bigger issue around AI and missiles, according to Manning, is the geopolitical uncertainty created from the fact that Russia and China cross nuclear weapons safety and security boundaries that the US sets for itself and its allies.
For example, US hypersonic missile systems like Dark Eagle are conventionally armed, while Russia and China’s are nuclear-capable. North Korea and India are testing the boundaries of US deterrence by developing nuclear-capable weapons as well.
These weapons are being developed at a much cheaper cost than US conventional missiles, not just because Washington spends a lot more on safety and oversight but because it pays for more parts of the research and development process like intellectual property rights, data repositories, and software.
In contrast, as authoritarian regimes, Beijing or Moscow can just demand new technology, including AI models, from their respective national enterprises for free to integrate into missile weapon systems. Other governments in Europe and Asia are likely to pool defence industries’ research efforts in a bid to keep up with the larger powers.
Interlinked missile barrages
One of example of battlefield technology developments involving non-US AI and missile technology comes from European consortium MBDA, which has developed the Orchestrike. This tool allows AI-driven co-ordination via missile-to-missile datalinks and specialist algorithms, that integrate missiles in flight with one another and the operator launching them.
The Orchestrike system allows missiles to react to threats and work together with the operator to solve tactical challenges and increase survivability and overall mission performance, while keeping human in the loop operability.
The company first demonstrated its system at the Paris Air Show 2023, and again last year at the Farnborough Air show, using AI-enabled Spear cruise missiles that operate collaboratively to strike targets in order of priority assigned by their operator.
In the longer term, AI-enhanced missiles will continue to grow in capability and become more autonomous, Boyd stated. Space too will certainly be considered a contested domain in future superpower conflicts featuring missile developments, with documented instances of ground-based missiles destroying satellites already.
Boyd forecasts that for the near term however, AI will be just one of many options for a warfighter to consider; and the victor in any near-future peer-to-peer contest is likely to be the side that can best manage the battleground complexity through judicious use of back-room AI functions, rather than AI-enhanced missile systems.
