Simona R. Soare
.
The UK’s Defence Artificial Intelligence Strategy is a blueprint for transforming the British defence establishment and its armed forces into an ‘effective, efficient, trusted and influential Defence organisation.’ The ambition is to embrace the advantages of artificial intelligence (AI) for defence, mitigate the risks, and make the UK a global leader in the responsible use of AI as part of a ‘once-in-a-generation’ defence modernisation.
Published 15 June, the strategy hits the right notes: it contains a clear threat and vulnerability assessment; proposes concrete solutions to build societal, political, and military support for AI adoption; and recognises how organisational transformation can drive successful innovation. But the strategy still needs to get the human talent and digital enablers right; manage expectations of specific deliverables; and set metrics for measuring progress.
The right approach
Some three years in the making, the strategy has four objectives:Transforming the UK armed forces into an ‘AI ready’ organisation by addressing human talent, policy, procedural and enabler challenges.
Adopting and exploiting AI to enable strategic and operational advantages, mainly through cross-domain integration, decision-making, and international cooperation with allies and partners.
Strengthening the UK’s defence AI ecosystem through cross-sector partnerships between government, industry and academia.
Shaping global AI developments to promote security, stability and democratic values and project influence. The document outlines a clear governance structure, designed to drive AI adoption in defence, that is to be embedded at higher levels of decision-making. Strategic coherence rests with two bodies. The first is the Defence AI and Autonomy Unit (DAU), which reports to the Deputy Chief of Defence and is responsible for defining defence-wide strategic and policy priorities in AI implementation. The second is the newly established Defence Artificial Intelligence Centre (DAIC), which reports to the Chief Information Officer and will act as the ‘technical coherence authority’. The DAIC will integrate the technical aspects of AI adoption. Avoiding the over bureaucratisation of the DAU and the DAIC will be important moving forward. Moreover, it will be essential to ensure both bodies have the right legal authority to coordinate and incentivise AI adoption across the defence enterprise. Whether this is currently the case remains unclear.
When it comes to AI adoption, the MoD maintains a balance between centralisation and delegated implementation. The AI adoption planning seems inspired by private sector practices where low-risk AI – or ‘AI Now’ meaning ‘mature data science, machine learning and advanced computational statistics techniques’ for business process automation, smart logistics and maintenance, and intelligence, surveillance and reconnaissance – is centrally monitored and implemented at unit level. Meanwhile, the higher risk and more operationally leaning ‘AI Next’, meaning ‘cutting-edge AI research and development integrated within our broader R&D pipelines’, is centrally directed.
The Defence AI Strategy will be supported by various initiatives designed to successfully deliver AI innovation. These include creating a pan-defence AI Technology Strategy, AI Practitioner’s Handbooks and an AI Concept Playbook to guide AI integration efforts.
The strategy acknowledges the importance of building new types of partnerships by harnessing the ‘triple helix’ model of AI innovation. This will involve:
establishing structures to facilitate cooperation, including a National Security AI Network to encourage broader engagement between and transparency within industry and academia
strengthening cooperation within the defence and security national AI ecosystem
enhancing transparency in relation to partners across the MoD on requirementsadapting processes to better facilitate cooperation, including through co-creation, targeted investment, and business growth support.
Meeting the challenge The success of the defence AI strategy will depend on several factors. An "AI ready" force must first get the human talent and digital enabling component right. The strategy lays out a vision of a new Defence AI Skills Framework and AI career development and progression pathways, including the possibility of lateral entry, personnel exchanges and increasing senior leadership’s understanding of the impact of AI.
The document goes into some detail about AI-related skill sets and career paths. It also discusses educational programs and how capability operators can be ‘AI ready’. However, the strategy has four major shortcomings.
Firstly, the strategy does not explain how its approach changes internal rewards in terms of career advancement for personnel with AI skills or for champions of high-risk, high-reward innovation. It also doesn’t address how it changes the organisational model of the Ministry of Defence in which the rotation of personnel into specific roles is often disjointed from their respective skillsets.
Secondly, managing expectations around the strategy’s deliverables in the short and medium term will be important. The strategy has high ambitions which depend on operational AI offering technological and decision-making advantages. In the short and medium term, the strategy’s implementation will prioritise consolidating the pillars of digital transformation (human talent and digital enablers) and adopting the low hanging AI fruit (or ‘AI Now’). The success of this approach depends on how the rollout of cloud and edge computing progresses. So far, progress has been slow and cumbersome. The US Department of Defence’s AI rollout has been delayed due to cloud computing issues, and the slow implementation of a pan-DoD data management system should be a cautionary tale for UK Defence.
Thirdly, the strategy does not have metrics for measuring progress. At the very least, such metrics should be included in the upcoming MoD-wide Implementation Plan of the Defence AI Strategy or in the ‘AI readiness’ guides for individual units. A lack of metrics will not only complicate monitoring progress but stop existing pockets of AI innovation in UK Defence from scaling up and building across domains and services.
Finally, the strategy makes no mention of resources whatsoever, which makes it difficult to determine whether science and technology and research and development investment is sufficient to support the level of ambition.
While challenges outlined above could impact the UK’s Defence AI Strategy, the strategy’s implementation is a step towards transforming the British defence establishment into an ‘AI ready’ force. Sustaining this momentum will now be key.
No comments:
Post a Comment