DAIC0016

Written Evidence Submitted by Professor Toby Breckon.

 

Views on developing AI capacity and expertise in UK Defence.

 

I am one of the UK’s most active mid-career academics in defence and security related AI research, specifically focussing on computer vision and autonomous sensing. My research has had significant industry and policy impact as I led one of the original research teams within the UK SAPIENT programme (gov.uk/sapient; now British Standard BSI Flex 355) and our work on algorithms for airport security scanners now is now used globally to help secure over 500+ million passenger journeys per annum. I current lead the Visual Computing research division at Durham University. I have acted as external scientific consultant both to industry and government on numerous occasions, and my PhD students founded COSMONiO which was acquired by Intel in 2020.

 

re: What strengths and expertise does UK industry currently have in the field of Artificial Intelligence with defence applications?

 

My view is that the UK has a strong research base that comprises the primary source of UK-based AI expertise and, based on the volume of UK scientific publications at top venues, punches significantly above its weight in terms of high-quality peer-reviewed research. This research base is currently spread across academia, a growing number of startup and scale-up AI-focussed businesses and also in UK-based AI labs of large ‘big-tech’ multi-nationals (e.g. Microsoft, Samsung, Toshiba, Google, Meta). In academia and industry alike, many leading researchers are based in the UK and globally the UK’s contribution to global progress in AI research is seen as significant. However, only a small portion of this research base is actively engaged in the defence and security field. Within academia, AI research engagement with defence and security is the exception and far from the norm whilst the large majority AI-focussed industry activity in the UK is not primarily focussed on supplying capability into the defence sector. AI expertise across traditional public and private sector organisations within the UK defence sector lags significantly behind academia and the broader UK AI industry in terms of the AI capacity and capability. Whilst UK-based academic and industry-led research is arguably world-leading in many areas, this world-leading capability is not defence sector based or defence application focussed.

 

As an academic, I see a significant reason for this relates to supply vs. demand in the UK AI talent pool. At present, AI expertise is very much in demand and the reward, flexibility and freedoms afforded to AI researchers in academia and (non-defence) industry research is often in sharp contrast to that on offer from the defence sector itself. My postgraduate qualified students often land well-paid industry jobs, based in city-centre business hubs with flexible working and the freedom to continue to publish their work, often focussed on fundamental AI research topics, in the open scientific literature. This latter point has increasingly become a culture of industry-based AI research labs – “yes, of course you can publish” – which allows them to attract top talent from academic roles, where publishing is the norm, and allows industry-based researchers to similarly develop their AI career via an increasing portfolio of openly published research. Labs which choose not to follow this culture, fail to attract the top talent in the same way in what is a very competitive jobs market. It is notable that many ‘big-tech’ multi-nationals specifically choose to locate one of their global AI research labs in the UK in order to gain access to this very talent pool – offering UK AI researchers global opportunities on their doorstep whilst at the same time increasing the domestic demand for UK AI talent. The challenge for the UK defence sector now faces is also compounded by the fact that a significant majority of UK based early-to-mid career AI experts are not UK nationals – making a career in the defence sector more challenging, and hence less attractive, on the basis of inherent security requirements.

 

re: How can the UK Government best develop capacity and expertise within domestic industry in sectors such as engineering and software to support the development and delivery of Artificial Intelligence applications in defence?

 

My view is that this comes down to the available future talent pool of AI researchers, and their availability to the UK defence sector. Whilst several public bodies are engaged with this challenge in the general sense (e.g. UKRI, EPSRC, DSIT - Office of AI, DfT CCAV), and there are specific efforts to increase career pathways into defence (e.g. EPSRC/DSTL University Defence Research Centre, DSTL Centre for Doctoral Training sponsorship, various HMG student internship programmes) it is clear to me that this is not going to deliver the required future AI expertise capability in the volume that will be needed by the UK defence sector going forward.

 

Whilst some efforts have been undertaken to increase the UK AI talent pool in general (e.g. targeted funding for AI-based Doctoral Training Centres), it is my observation that only a relatively small number of UK computer science (software) graduates enter the UK defence sector, both at undergraduate and at postgraduate level.

 

What is required is a large-scale, sector-wide efforts to increase the volume, scope and diversity of AI researchers available as a talent pool to the UK defence sector and increase the attractiveness of an “AI in defence” career when placed side-by-side with other career paths available across the UK AI industry. At the entry-point, this could be via increased financial support incentives for students studying AI at UK universities entering the defence sector and increased support for defence-focussed PhD training opportunities in AI. Furthermore, facilitating mid-career up-skilling of defence sector personnel through funded study-breaks (as are more common overseas from my experience) or increased incentivisation and opportunity for academia-industry collaboration would also aid in growing UK AI expertise capacity in the sector. Longer term a culture shift maybe required such that both public and private defence sector entities can begin to offer a long-term attractive career path to the broader AI talent pool in the UK.

 

I believe that making the UK defence sector a more attractive place for AI researchers to work, against a backdrop of fundamental long-term research in AI aligned to defence sector requirements, will enable UK defence to leverage the world-leading AI research capability that the UK already holds as a nation.

 

 

 

17th February 2024