From trials to frontline capability: Accelerating AI adoption in UK defence

Jack Maxton is CEO of DIGI2AL and a defence technology expert with experience across government and industry. He previously served as a Chief Digital & Information Officer at the Ministry of Defence and has held senior technology roles at Credera, KPMG and the UK Civil Service.

We’ve recently seen a great deal of handwringing about the efficacy and risks of AI. While little of that commentary has focused on defence (so far…), both allies and adversaries are accelerating their use of AI in defence.

We must respond, not just to keep up, but to lead in this area, exploring AI’s use and deployment in operational scenarios to improve mission outcomes. How can that be achieved?

The Defence AI Centre’s role in UK military innovation

The Ministry of Defence is clearly taking AI seriously, indicated by the establishment of the Defence AI Centre (DAIC), whose charter is to “champion, enable and innovate artificial intelligence across UK Defence…for the strategic advantage of our Armed Forces.”

DAIC has already made some big strides by building strong partnership networks, running evaluations and defining policy.

While it’s tempting to try to assess whether we’re behind or ahead of the curve, what really matters is that we keep innovating and spend a little less energy striving for perfect, risk-free deployment and more time putting theory into practice.

That doesn’t mean making risky choices; it just means trying things in the real world and learning more quickly what works and what doesn’t; the challenge is as much attitudinal as it is technical, and adopting this approach will keep us at the forefront.

AI opportunities in MOD productivity and cyber security

There are in fact several simple, low-risk areas where we can push ahead and gain significant benefit in relatively short order. Chief among them is productivity through better administration within the MOD.

For example, simple tools like Copilot can be used for productivity scenarios such as note-taking and summaries of meetings, which can reduce required attendance and speed up actions, releasing staff from repetitive or laborious tasks to spend time on more productive matters.

Another example is cyber security of the MOD itself, dealing with the vast array of low-complexity cyber threats the MOD faces every day. AI can be used to improve attack detection and mitigation, again speeding up resolution and freeing up resources.

Both are examples where adoption is relatively straightforward and fast but are also very low-risk with very high return. A somewhat more sophisticated use of AI is in intelligence and operational environments, for example, image and object recognition and operational digital targeting, but with a human in the loop to provide error checks.

Integrating AI into defence digital transformation

Examples aside, how do we put this into practice? Both from philosophical and practical perspectives, we have to stop thinking about AI as a standalone solution, separate to general digital transformation initiatives.

Instead, we should think about AI as a key element to be integrated into a bigger system or project that drives operational benefits, weaving in AI strands and pushing AI’s capabilities from the outset.

However, this is where we hit a bit of a barrier. The digital skills base across the MOD, with free time and capability to experiment and implement AI programmes, is relatively low. While the enthusiasm is there, the fragmentation and maturity of AI experience across the department often means it is easier to give up and return to the day job than it is to build a coherent strategy and capability.

Industry collaboration and military AI investment

An industry-defining technology shift can’t be the sole responsibility of a single institution. Fortunately, DAIC is already demonstrating the value of building closer ties with the UK’s industrial base. Normally acting with fewer regulatory constraints, these organisations are used to building, testing, failing fast and building again.

They are able to learn quickly from pilot projects and move to minimum viable products (MVPs) and solutions much faster and further than the MOD can do currently. These organisations already have the solution backbone and tools ready to experiment inside the defence tent.

It’s therefore clear that we now must work harder to get these organisations on board to bring in new thinking, new experience and new technology to defence AI development. The government’s recent Defence Investment Plan sets aside £300 million in dedicated military AI funding, covering land, sea and air domains, intended to attract and engage a wider supplier base.

But while the effort will be worth it in the long run, there is perhaps a sense that this is a relatively small budget for such a seismic technology shift, a budget that will be quickly swallowed up and spent.

And so we return to the discussion about efficiency, and how that budget is best used. It’s critical that new players can access the defence ecosystem to help it help itself, but it’s equally critical that the MOD is open to new, creative ideas and practices.

While the rest of the world is already adopting AI, our opportunity lies in flexibility, openness and a willingness to start adopting AI at speed through trials and prototypes.

Moving from AI trials to operational adoption

As we build our learning and experience, we can begin to think about mass adoption, moving from innovation projects and evaluations to full integration into live systems, albeit with the usual considerations around operational risk. As mentioned earlier, we can start to realise the benefits of some quick wins now but that wider adoption will drive real operational advantage for digital warfighters.

These ‘build, test, refine, deploy’ approaches are not new for us at DIGI2AL, but are perhaps still not widely adopted across the MOD. We now need a collective, attitudinal mind shift across government, departments and industry that embraces this ethos. Our success with AI means not just having a few successful AI trials, but true adoption that keeps us ahead of our adversaries and leads the way with our allies.