Developing AI Capacity and Expertise in UK Defence

Judith Cummins Excerpts

Monday 13th January 2025 Commons Chamber

Select Committee statement

[Madam Deputy Speaker (Judith Cummins)]

We now come to the Select Committee statement on behalf of the Defence Committee. Mrs Emma Lewell-Buck will speak for up to 10 minutes, during which no interventions may be taken. At the conclusion of her statement, I will call Members to ask questions on the subject of the statement. These should be brief questions, not full speeches. I emphasise that questions should be directed to the Select Committee member, not the relevant Minister. Front Benchers may also take part in the questioning. I call Mrs Emma Lewell-Buck.

[Mrs Emma Lewell-Buck (South Shields) (Lab)]

Thank you very much, Madam Deputy Speaker. It is a pleasure to present this report on behalf of the Defence Committee.

It is important at the outset to say that our report is not focused on the ethical questions that using artificial intelligence in defence raises. That is because noble Lords in the other place completed a very in-depth Committee report on AI in weapon systems prior to our inquiry. In that report, they rightly addressed issues of safety, compliance with international law and the need for a human in the loop. It is also important to say that the key defence documents we refer to throughout our report were produced under the previous Government. However, we have not received any indication from the current Government that they will be making any significant or major departures from those documents.

Our focus was on trying to understand what the Ministry of Defence is doing in this area and how it can develop a successful defence AI sector in the UK. Since 2018, AI has been appearing increasingly in defence documents, as well as being spoken about by Ministers and senior military figures. The defence artificial intelligence and autonomy unit and the defence AI centre have since come into being, and the defence artificial intelligence strategy was published in 2022. Our strong research pedigree in computer science and the availability of computing power could put the UK at a strong advantage, and pillar 2 of our AUKUS partnership continues to present opportunities for working with our allies and for interoperability.

Despite that, we found that, overall, there is a “say-do gap” where the Department is identifying the right priorities, but has been unable to say what steps it will take to achieve them. Rhetoric does not match reality. This leaves us behind our allies and disadvantaged when it comes to our adversaries. Similarly, we consider the Department’s aim to be AI-ready problematic. Technology is evolving at such a rapid pace, and the MOD will never be sufficiently AI-ready. Instead, it should aim to be “AI-native”, with AI viewed no longer as a niche add-on, but as a core component running right through defence. This will take leadership, cultural and practical changes.

AI is a general-purpose technology, and although we do not have a clear idea of the size of our defence AI sector, we do know that it is young and underdeveloped, and that many of the companies that will supply defence in the future are likely to cut across the civilian and military sectors. For the MOD to give to these less traditional companies and investors any confidence and incentivise them to do business with the MOD, it needs to be clear about what it wants. It needs to show more flexibility, be less prescriptive in requirements, address delays in security clearances and fully embrace the recommendation from the Haythornthwaite review about how the brightest minds can zig-zag between the military and the commercial civilian sectors.

On new capabilities, the MOD is not comfortable with rapid change, risk or experimentation, but that is exactly how our tech companies and start-ups operate. The MOD’s existing approach of contracting with primes and spending 10 to 15 years building a ship, tank or aircraft to exquisite specifications just does not cut it in an increasingly data and software-driven environment. It is welcome that the global combat air programme is being designed with open-system architecture, meaning that software capabilities can be plugged into and taken out of the hardware, and the new spiral model of overall defence procurement in the integrated procurement model is welcome, too. However, our Committee found no evidence of fundamental changes across the organisation. If smaller, innovative companies are to engage with the MOD, the MOD must be a better customer.

Key to developing our defence AI ecosystem is the digital infrastructure, yet the Department’s plan for a digital backbone is behind schedule. Data collection, labelling and sharing, which are all key to training defence AI models, must be improved. Clarity of leadership and responsibility is also needed. The defence artificial intelligence centre is headed by a one-star officer, and we remain unclear on where the centre is based. There is fragmentation across the MOD, as the frontline commands of each service have individual responsibility for AI development, which causes confusion and duplication.

Ukraine has shown us that AI is changing the battlefield. AI has the potential to transform UK defence, from the back office to the frontline. In such a rapidly developing environment, we cannot predict what will come next, but we know for certain that our allies and adversaries are using and developing AI for military advantage. We remain hopeful that through the strategic defence review, and the Government’s focus today on AI, and by adopting our recommendations, the Department will start to turn its rhetoric into reality, so that it does more than it says.

I want to acknowledge what a team effort producing this report has been. I thank the previous Committee; many of its members became firm friends of mine, despite our political differences, and I dearly miss them. I also thank the new Committee, which has worked very hard to finalise this report. I thank everyone who has engaged with us, including the MOD and the previous Minister, the hon. Member for South Suffolk (James Cartlidge), who is on the Opposition Front Bench. I also thank our excellent Committee staff, especially George James, who managed the inquiry, and Professor Kenneth Payne, our very knowledgeable expert adviser to the inquiry. I look forward to taking any questions from right hon. and hon. Members, and I sincerely hope that they like the answers that they get.

[Madam Deputy Speaker (Judith Cummins)]

I call the shadow Minister.

--- Later in debate ---

[Mrs Lewell-Buck]

I could not agree with the hon. Member more. A section in our report addresses in detail SMEs and their engagement with the Department, and we should carry forward the work that he did in government, especially on the integrated procurement model.

[Madam Deputy Speaker (Judith Cummins)]

I call the Chair of the Defence Committee.

[Mr Tanmanjeet Singh Dhesi (Slough) (Lab)]

I place on record my gratitude to my hon. Friend for chairing the Sub-Committee, and I thank members of the previous Defence Committee, and Committee staff, for all their hard work on our report on artificial intelligence—a matter integral to our nation’s defence. While I welcome the Prime Minister’s announcement today about embracing AI in government, does my hon. Friend agree that there seems to be a chasm between the Ministry of Defence’s rhetoric and the reality? In practice, AI is still treated as a novelty, rather than a fundamental part of the MOD toolkit.