Digitise or Die? The Defence Digitisation Dilemma...

 

The Public Accounts Committee has issued a damning report on the MOD efforts to digitalise, stating that the Department must fundamentally change the way it is trying to deliver digital transformation. While not a hugely glamorous subject, nor one that attracts much attention online when you could be arguing over how many anti-ship missiles a Batch 2 RIVER class should have (the correct answer is, of course, none), this is probably one of the most critical parts of the defence portfolio. Failure to get this right could leave the armed forces trailing behind their peers and foes.

At its heart digital transformation is about taking the some 2000 IT systems used by the MOD and by 2025 achieving three core goals – the creation of a ‘digital backbone’ that can effectively share and use data across the MOD, the development of a ‘digital foundry’ to ensure the Department can properly use and exploit this data, and finally the creation of a workforce of digital specialists who can use it to its fullest capability. To give some sense of scale, the MOD spends roughly £4.4bn per year on digital initiatives, almost 10% of the Defence budget.

The problem is that things aren’t working out as planned. The PAC report walks through how the MOD accepts that data is transforming how warfare is conducted, but it doesn’t yet know how to properly use this to best effect. The MOD apparently treats the acquisition of digital projects in the same way as conventional military equipment, subject to the same processes and procedures. These may be fine if you’re looking to buy a tank or warship, but less effective or speedy in the rapidly changing world of software. The other key challenge is that the MOD doesn’t really have a plan to coherently track delivery and integration of these projects in a meaningful way – in other words, over £4bn of public money is spent each year without there being a single master project plan to track how it all comes together.

These issues manifest themselves in a variety of ways – right now for example the MOD admits that it doesn’t know what data it holds about different areas. It also makes processes extremely clunky and laborious – trying to order a pair of combat boots is cited as a particularly complex task due to the clunky legacy infrastructure in place. This is the problem – the MOD has, like all modern armed forces, had to adapt to an almost terrifying speed of change in the digital space in the last few decades. The lifecycle of equipment and infrastructure is far longer than the speed of IT change – consequently the MOD has to balance of over 2000 different systems which entered service over an extended time frame, and which need to be able to communicate with each other and work.

The scale of the MOD digital task is almost mind-blowingly complex – as a globally facing organisation employing over 200,000 people operating across all seven continents and oceans, and from outer space to deep underwater, the MOD needs to operate, maintain and protect an incredibly complicated mixture of digital systems. This ranges from the low end – for example logistics systems and store inventories to cash and payroll systems, to maintenance of IT networks including battlefield communications, highly classified securely protected IT systems able to operate at up to TOP SECRET, and the ability  to communicate globally between different locations. In extremis the MOD needs the ability to ensure that the Prime Minister can request a nuclear strike and send this via a protected nuclear firing chain. To try and get your head around how all of this works is baffling, but the upshot is that none of this was constructed according to an ideal plan of sticking it all together. Instead the system has been brought about through luck, judgement and procurement at different times over many decades.

The challenge now is to get a system in place that works, and its clear that this isn’t really working out. The risks are huge -if the MOD can’t get its digital transformation right then not only may combat boots not be able to be ordered, but more serious issues like whether Bowman remains fit for purpose or could be compromised emerge. The Ajax and other advanced vehicles in the pipeline may not be able to operate to their full and intended potential if they don’t get the full suite of digital upgrades needed to communicate properly. If this happens the multi-billion pound vehicle will be fundamentally useless as it cannot be used as a networked platform able to embrace modern technology. This is just one example of how getting this wrong can potentially hold huge risks for the armed forces and have a direct impact on the battlefield.

The Army take on Digitialisation - Image by Ministry of Defence; © Crown copyright 2023.

The other problem is that the MOD doesn’t know how to properly use the data that it does have – for example can it fully analyse and assess what it means and use it to its fullest advantage. How does it take advantage of the myriad of data sets flowing through the system and help be more efficient and able to exploit this for commercial or military advantage? Part of this comes from recruiting the right people able to understand the systems, use them properly and provide advice that can be used. This in turn requires recruiting a digitally skilled workforce able to fully exploit the systems they have, which the report notes is something that the MOD is failing at.

Part of the problem is the same old story that Civil Service salaries are simply not keeping up with the civilian competition, leading to a reliance on external consultants and contractors. There is a growing skills shortage in the MOD which is exacerbated by the need to recruit in the Corsham area, near Bath, where the digital work is led from. The report notes that there is no spare talent to draw on now (particularly as DE&S also recruit locally), meaning that finding people who can afford to live in the expensive Bath area and are willing to realistically take a significant pay cut to join the Civil Service are in short supply.

This highlights part of the problem of regionalising the Civil Service outside of London. While there are many very compelling reasons to move the CS away from Whitehall, the challenge if you fix operations on one geographic area and need highly skilled roles is that you quickly run out of people. If you want to run digital operations in one space and refuse to pay moving expenses, and expect people to move to an expensive area, it shouldn’t come as a surprise that people don’t join. This also highlights the reason why, despite several MP’s offensively suggesting that all the CS do is work at home, there are huge benefits to flexible working – if you can recruit a national workforce and let them work remotely, you stand far better chance of getting people to join than you do if insisting people physically come into office. This strikes at the heart of the ‘work remotely or work at office’ debate – what matters more – people in an underrecruited office without the skilled workers, or a remote workforce with the right skill set?

The wider challenge against delivery of all of these initiatives is that the MOD is financially in a very difficult place right now. Inflationary pressures have significantly reduced the value of the defence budget, and the UK is alone among all major NATO nations in not committing to short term cash increases to the defence budget as a response to Ukraine. This means very difficult decisions loom about what to prioritise and what to delay, descope or cancel. Given the scale of the digital portfolio it is likely real challenges will be faced in prioritising what to keep and what to scrap or reign in, and this will have a knock on impact on operational capability. Very few people would want, for example, to see warships scrapped over preserving funding for a new IT network, yet in the medium term if this network isn’t funded, the operational effectiveness and survivability of these same ships may be drastically reduced. Ministers and senior officers face painful and very difficult decisions to balance budgets that will have a negative effect regardless of choice on the armed forces.

One area where digital infrastructure is becoming particularly interesting is the growing use of AI in different forms. The ChatGPT is a so-called ‘chatbot’ that can provide fascinating and good written answers to questions, that in turn is raising all manner of ethical concerns around its employment. Some schools, for example, are reviewing their homework policy as essays written using ChatGPT are indistinguishable from human ones – pupils simply set the task to the AI and it is done for them. For the armed forces ChatGPT could potentially pose a terminal threat to the use of written reports for the annual promotion cycle.

Currently in the UK promotion in the armed forces is conducted through an annual reporting cycle. Everyone has a report written on them by their superior which comments on their performance and uses evidence to show how, on merit, they are suitable for promotion. There is a space for a second (and sometimes third) reporting officer up the chain to enter comments too in certain circumstances. In each case the report is designed to show how the person is progressing, whether they are deemed suitable for promotion and enables them to be assessed against their peer group at annual promotion rounds. It is a system that is ‘unique’ and arguably increasingly unfit for purpose.

The reporting system relies on a good author with knowledge of the individual – not always possible in the world of regular moves to new posts. Where people have moved, the timely submission of written analysis on their performance is essential to ensure they get the best possible chances of an accurate and fair report. The author knows far too many people who have given their all each year, only to be seen off by either a reporting chain not bothering to write or submit an accurate report, or for comments to be stuck in a JPA workflow from hell, preventing them from being seen. While of course a Service Complaint is an option, it doesn’t take away that the person has missed out again on promotion for another year.  In particular the quality of writing is often highly variable – to write well and accurately is a skill, particularly for the OJAR/SJAR process. It is entirely possible for hugely capable and qualified individuals to not promote because despite a strong performance, the author of the report did not quantify well enough why promotion was appropriate on merit. How many good people have lost out because their superior didn’t do their job properly?

This raises the question as to whether ChatGPT is the answer – feed the chatbot with sufficient data points and evidence and let it draft a report for AB Bloggs that will show to a high written standard why they should be promoted. It raises a very interesting line of debate – some busy or less than scrupulous officers may feel that using an AI bot to draft a report saves time and makes their lives easier, and also produces a strong report. Promotion Boards are unlikely to be able to have the time to tell whether a report is drafted by a person or an AI bot, which means it could rapidly become the tool by which most OJAR/SJAR drafting is done by some people. For those officers who adhere to ethics and values, this places them in a bind – should they continue to draft their own report, even if it is less effective in doing the best it could do in selling an individuals performance and maximising their chances of success at the promotion board? Or should they too use Chat GPT to level the playing field and ensure everyone has the same chance of success?

Early uses of this system have produced mixed results- reasonably good first drafts but lacking in evidenced detail. But the system learns and evolves. Its not hard to envisage circumstances in the near future where the rapid rise of AI Bots able to draft highly accurate evidence based reports, particularly as they learn style and feedback, will make it hard to hold a promotion board which can objectively assess its subjects. There is also the wider question of if the promotion process exists to show someone’s performance objectively, and the AI Bot has the means of doing this, is it better to use an impartial system to objectively display performance narrative rather than rely on a reporting officer whose skills and written ability may be weak? What is better for the long term needs of the Service – is it to write clunkily but personally and see good people miss out, or rely on AI and automate the process to a high degree?

Its likely that this debate will continue for some time to come, but it is hard to overstate just how big a deal this poses for the long term future of the military reporting system as its currently constructed.

While the days of killer robots and Skynet are (one hopes) going to remain in the realm of science fiction, the fact is that digital transformation, AI and other initiatives is changing how the armed forces will work at lightning speed. The current system of procurement and approach isn’t fast enough to embrace this, and the right people aren’t necessarily being brought into deliver it. Short of rapid change and a very different approach to embracing the digital future, the MOD may find itself left behind and struggling to catch up. It seems a fairly binary choice on what to do next…