Building an Artificial Defence

Chief Data Scientist, BAE Systems Applied Intelligence Read time: 4 mins
Although there is clear potential for Artificial Intelligence in Defence, it’s important not to get ahead of ourselves. David Henstock explains why it’s more marathon than sprint.
Building an Artificial Defence When I’m asked about Artificial Intelligence (AI) and Defence the questions come thick and fast from people who don’t actually work in the sector. Their interest has often been sparked by Hollywood blockbusters which portray a future shaped by military machines with a life of their own.

But while the hype machine has been in overdrive, expectations also need to be realistic. We’re quite a long way from AI and machine learning gaining ascendancy – in Defence and elsewhere. That said, it is also clear that such technologies could potentially transform military operations – starting sooner than you might think.

Momentum building

I am often speaking to groups of people about data, technology, and transformation. It’s been noticeable, however, how such talks have increasingly been to those working in and around Defence – and this has been driven by a number of critical factors.

This surge of interest has been underpinned by a convergence in requirements, computer processing power, and clever technology. Defence is becoming increasingly focused on exploiting data but it can be hard for military personnel to cope with its volume or speed on a modern battlefield. Processing platforms have also evolved to provide the necessary scalability and there is a rich ecosystem of AI technology development originating from outside the Defence world – for example in finance and healthcare – that Defence can take advantage of.

AI can benefit Defence in a number of ways. It can assist the military user to make decisions – preventative maintenance, supply chain optimisation and combat decision support are just a few examples which spring to mind. But AI can also drive completely automated decision-making in certain areas that augment and improve human capabilities as well as providing additional protection for our armed forces. We are starting to see some of this with autonomous vehicles, and may in the future see automated response to threats.

But listing the benefits is one thing, actually achieving them is quite another.

Lessons learned

To get the best out of AI and machine learning, the military need to change and open up their procurement model. Defence organisations should court individuals and groups such as those in academia or cutting-edge start-ups conducting pioneering work and building open source software,  and consider how they can deliver individual AI components and services. The drive for new AI technologies and applications will only bear fruit if both collaboration and agility become fundamental to any approach.

But it’s not just about innovation. Of equal importance are the practicalities. What happens to the innovative ideas and proposals? For the military, it is about operationalising capabilities and this means a change of mindset is required.

Some of this is focusing on the data that will be used by AI: where does it come from, and how should it be managed and integrated? But it is also about successfully deploying and maintaining AI capabilities in a dynamic, challenging and uncertain environment, and how you help military users engage with and trust AI. While research is currently at the forefront of activities, the operational implications need to take centre-stage if the technology is to truly  succeed in military theatres.

And even if this change of approach takes root, that’s no guarantee of success. Challenges such as attracting skilled AI engineers – not easy when higher salaries are on offer elsewhere – and understanding how to deal with the obvious ethical issues remain ongoing.

Eyes on the prize

In the short term it seems likely that AI in Defence will be about specific solutions to specific problems. Although senior officials and engineers may yearn for a faster pace, it will be a journey built up over time. But in the meantime, Defence should invest in the enablers by focusing hard on agile delivery and innovation collaboration across academia, SMEs and mainstream defence suppliers.

The key priority ahead, though, is operationalising the technology. The transfer from scientific laboratory to military theatre is pockmarked by potential barriers but by addressing issues surrounding data, trust and the practicalities of deployment, Defence organisations – just like their counterparts in other parts of the public and private sectors – can turn AI ambition into practical, long-lasting impact.

Discover the latest trends, topics and technologies; explore more Government Insights from our experts.

About the author
David Henstock is Chief Data Scientist at BAE Systems Applied Intelligence

David Henstock Chief Data Scientist, BAE Systems Applied Intelligence 31 October 2019