Adopting AI in Defence

Introduction

There has been lots of talk around AI in the UK Government in recent days, the PM’s speech gave me great confidence we’re planning on orientating our government to get after the value AI brings. However we’re missing the other side of AI which is critical to the efforts to build and maintain a modern armed forces. A lot of AI investment in Defence is focused on the hardware and operational side and ?will ensure we maintain or exceed parity with our adversaries around the world. Additional focus is needed to leverage AI for the everyday which is where we are feeling considerable pain.

Widening the scope to focus on the value

In Defence, we need to widen how we view and leverage AI.

Its not just a tool to parse EW signal data, or to combine different data sets to aid with operational planning. AI is an enabler of both improved productivity but also organisational agility. When we look at the current economic conditions and the pressure they are exerting on budgets we can predict a certain amount of turbulence within the Armed Forces.

We are left with two conflicting challenges; firstly, when we change following a review or fiscal event, we will lose vital experience which leads to a drop of quality and productivity. Secondly, our budgets are being tightened so we need to improve productivity without losing the value of what we are doing today. To say we’re stuck between a rock and a hard place is an understatement.

AI can play a critical part in this process when we get down to human machine teaming. The Rt Hon Peter Kyle MP highlights this in a number of places in the 50 point plan. References to leveraging AI Assistants is a critical and often a previously overlooked point, particularly because its not always considered the Gucci Sexy use of AI.

AI for Everyone, Every Day

Productivity improvement through AI Assistants are vital as we continue to strive to do more with less, automating the repetitive low value high volume work is crucial to both improving productivity but also to facilitate greater organisational flexibility without losing the learnings and experience held by individuals performing the work today.

AI Assistants can play a vital part in this but we also need to expand our horizons and look at how AI can be embedded in not only assistants but the work our armed forces perform on a day to day basis.

AI needs to become common place in supporting the processes and decisions we make, if that’s predicting when a part will break so replace it before it does (rather than on operation when its too late) to evaluating and monitoring risks in our compliance and audit processes.

Both these examples point to the same challenge; embedding AI is not just about automation its about enabling better and faster decision making in every day work.

A lot of the AI tools today enable users to access a model and extrapolate some learnings to apply to their work. We need to take this to the next level and integrate the models directly in the processes themselves and not just leave at another piece of kit in the stores to break out only when needed. AI should become more than a tool to refer to on occasion, it should be sitting on our shoulders and leveraged directly in our processes to coach, guide and prioritise our work.

Rapid AI and Rapid Value

My final consideration when thinking about the 50 recommendations refers to the SCAN > PILOT > SCALE methodology.

I am incredibly positive about this approach as it lends itself to a truly agile and iterative approach to leveraging AI. AI models and tools are evolving at an astounding pace and without a proper approach and tool set, we risk falling behind the latest innovations before we complete our current set of projects.

Being able to rapidly prototype and pilot tools and applications is vital, whether they leverage AI or not. We all know how long innovations takes to reach the hands of the users and addressing this problem is not just an aspiration but a necessity if we want to maintain our technologic parity.

There are a huge number of tools available today which facilitate this, from everyday uses such as Generative AI to create process documentation to other generative models that allow you to parse documentation into design documentation and then into a working software application.

Often these tools leverage historical insight into how similar problems have been solved before, this allows individuals to both step outside their knowledge areas and then to quickly learn practices techniques to apply to the problems they solve.

Finally the framework and platforms we use to build out our IT infrastructure should be considered. Adopting low code platforms that leverage AI at their core allows the Defence ecosystem to maintain operational agility and continuously adapt our processes as models and learnings improve. Traditional high code development simply cannot maintain pace in the current environment and we need to focus on shifting away from this as a common practice to high code development by exception only.

In Conclusion

I’ll conclude my rather rambling thought piece on AI by saying the 50 point plan contains a lot of valuable directions and I sincerely hope we achieve the ambitions set out.

Critical to this will be ensuring we fully appreciate the scope AI can deliver on and shift our thinking towards addressing the full range of problems that AI can solve in Defence.

Adopting AI is not just about the models we leverage it is critical to understand how we can get it in the hands of the every day user and avoid it becoming just another tool on the shelf.

Finally, agility in transformation has always been an ambition in Defence and to meet the increasing pace of change we will need to adopt low code platforms and frameworks.