In a recent press release, the UK Government states that 20,000 civil servants used Microsoft 365 Copilot for a variety of tasks, including drafting documents, summarising meetings and more innovative use cases such as personalising advice for jobseekers.
Overall, the Government estimates that two weeks could be saved annually per civil servant, with an aggregate impact equivalent of giving 1,130 people a full year back.
Compared with the types of use-case and indicative time savings identified elsewhere in the public sector (for example in the Local Government Association AI Case Studies), these encouraging figures appear perfectly reasonable.
The Government hails this as part of its effort to “modernise the state and achieve £45bn in savings.”
A rough estimate suggests the value of a year of time for 1,130 civil servants could be in the order of around £50m. If this were expanded to the whole civil service, as the press release suggests it could be, the annual value could be approximating £1.25-1.5bn Not enough in itself to achieve the Government’s target, but a very healthy start.
Here’s the catch
However there a couple of catches.
US Tariffs are shifting - will you react or anticipate?
Don’t let policy changes catch you off guard. Stay proactive with real-time data and expert analysis.
By GlobalDataThe first is that Microsoft 365 Copilot isn’t free, nor included in any existing enterprise subscription. At the time of writing, the annual list price is an additional £277.20 per user in the UK. So, the usual annual cost for 20,000 civil servants would normally be just over £5.5m. For the whole civil service (a little over 500,000 people) it would be around £140m.
While still offering a very impressive approximately 10x return on investment, this is serious money. And of course, there is nothing to stop Microsoft increasing its subscription charges in the future (all this glosses over the relative costs and savings for part-time staff and the fact that the Government could probably get about a 20% discount on Copilot due to the existing Strategic Partnership Arrangement 2024).
And there’s more
But this isn’t the biggest catch. The most crucial point, conveniently overlooked in the Government press release and much reporting of it, is that none of that £50m value in the trial is a cashable saving.
As GlobalData Principal Analyst, Gavin Sneddon, observes, “Time saved is added value. It is not a saving. It only becomes a saving if it is accompanied by a staff restructuring which makes people redundant. And this is much harder to do than simply investing in new technology.”
In order for Copilot to contribute to Government savings targets, it needs to enable the shedding of staff, and nothing in the trial use cases makes it obvious how this might be done. So as things stand, the Government is looking at spending significantly more money to create extra capacity, rather than realising any savings.
Savings real or imagined
It might be pointed out that this is just one small strand of the Government’s digital ambitions, with the main savings to be realised by the automation of routine activity.
After all, didn’t Politico recently lead with the promising headline that “The UK government thinks AI can do two-thirds of the most junior civil servants’ work”? Well, it certainly did, but unfortunately a tradition has arisen in journalism in which headline writers are divorced from article writers, and the Politico article says something quite different.
In fact, it notes that the Government “hadn’t assessed whether existing AI could actually automate routine tasks.” This is based on a methodology note, dragged kicking and screaming from DSIT which states that “This analysis solely considers the routine nature of tasks and does not assess the current feasibility of automating them using existing AI capabilities.” Let that sink in.
Cuts to bring savings
Of course, separately, the Government does, in fact, have plans to shed civil service staff and save money. Rachel Reeves has previously said that the Government is confident it can “reduce civil service numbers by 10,000”.
Departments have been set ambitious targets to cut their annual revenue spending of between 10-15%. But, while Reeves has made vague gestures to the cost-saving potential of technology, this process does not appear to be integrated in any meaningful sense into the technological initiatives under way. At present the Treasury and DSIT each appear to be dancing to their own particular tunes, to say nothing of individual government departments.
The caveats in section 7 of DSITs methodology note are incisive and succinct (and entirely vindicate the analysis in a previous piece from Verdict on this subject): “…given the system challenges of delivering in the public sector, specific implementation assessments should be at a local level – factoring the current digital maturity of each organisation.” But so far as we can tell, absolutely none of this is, as yet, under way.
So what is likely to happen? Sneddon suggests that “In these situations organisations typically make cuts and then hope that, magically, technology will take the strain. While this could actually work reasonably well in some services, others are likely to find that staff-cuts are not offset by the extra capacity provided by new technology.”
Whichever way it unfolds, in the absence of an integrated research and development (R&D) and workforce strategy, nothing in the recent announcements suggests the Government will actually make much direct headway towards its £45bn savings aspiration. The technology servant is looking encouragingly well-suited, but the change-leadership emperor appears to remain as stubbornly naked as ever.
