Authorities wants to repair tech workers hiring course of, says Whitehall’s AI boss

The Science, Innovation and Expertise Committee has questioned tech specialists on the federal government’s plans for the digital centre of presidency – or the brand new Authorities Digital Service (GDS).

The committee examined the position of the brand new centre and whether or not the federal government’s digitisation of public providers will likely be profitable.

Earlier this month, prime minister Keir Starmer promised a possible £45bn in financial savings ought to the federal government get digital providers proper, and pledged to ship groups into authorities departments to make sure they’re utilising know-how and synthetic intelligence (AI).

That is a part of the federal government’s blueprint for digital authorities, revealed in January 2025, which guarantees agile know-how supply, upskilling of civil servants and a concentrate on AI.

Giving proof to the committee, the federal government’s head of AI, Laura Gilbert, mentioned there are at present lots of people in senior digital, information and know-how (DDaT) roles in authorities who she wouldn’t “contemplate know-how folks”, including, “I wouldn’t rent them”.

“It’s very troublesome to rent technologists properly. The way in which the civil service hires for this type of position will not be appropriate for this type of objective. It’s actually hit or miss,” she mentioned, including that having CVs with the proper buzzwords and programming languages checked off doesn’t essentially make these folks the proper match for the job.

“The system doesn’t have a solution to rent that assures that individuals [can] do the job. Till that adjustments, it’s going to proceed to be very troublesome.”

She added that to turn out to be an AI engineer within the authorities’s AI incubator, she places potential hires via a four-hour coding take a look at, testing their abilities in an actual state of affairs, whereas the civil service “remains to be doing behaviour-based interviews and attempting to use that to technical roles”.

Gilbert mentioned she can be involved concerning the prime minister’s announcement that, by 2030, one in 10 civil servants will likely be in a digital position.

“Will they actually be digital abilities? Would Google rent them? As a result of that’s the usual we must be going for,” she mentioned.

The committee additionally questioned the potential £45bn in financial savings, a determine that comes from a Bain & Firm report revealed in January 2025.

Gilbert mentioned the methodology for developing with these numbers “is completely good” however, as with every little thing else, it’s guesswork.

She mentioned there’s a front-loaded half the place there are straightforward positive factors on fairly huge initiatives you’ll be able to spin up that don’t depend on holding legacy methods afloat, however a variety of it is going to be long-term initiatives that want folks and cash, and he or she doubts the federal government will ever get to a degree the place they will say they’ve saved precisely £45bn.

Former civil servant turned guide Richard Pope, director of Richard Pope and Companions, added that it’s “actually arduous to digitise an analogue service and never lower your expenses”, so there’ll naturally be financial savings.

“However initiatives predicated on saving cash have a tendency to not work very properly for the general public,” he added.

He’s additionally involved that the federal government has publicly targeted an excessive amount of on AI because the saviour.

“I ponder if ministers have painted themselves into the nook with AI a bit of bit, as a result of it’s nearly like AI has turn out to be synonymous with financial savings,” he mentioned, including that there are different digital merchandise, providers and methods that won’t sound as flashy, however work properly, and that to attain the financial savings predicted, the federal government wants to make use of “the complete scope of digital”.

Each Pope and Gilbert are involved by the dearth of readability round what the federal government is attempting to do in information trade.

Pope mentioned that previously 5 years, GDS has completed 5 totally different information trade initiatives, and the federal government’s blueprint alerts that one other one is within the works.

Gilbert is anxious that the worth across the information programmes thus far has been “actually low and never far-reaching”. She says the plans across the information library and information sharing have to be “actually clear” on why it’s being completed and what drawback it’s attempting to unravel, whether or not that’s information trade inside authorities, public information, private information, what the outcomes will likely be, and the place the cash is being spent.

“The sensible implementation is essential. There’s a actual lack of readability on what persons are attempting to do within the system,” she mentioned.

Pope added that the brand new GDS wants to indicate that it could possibly train affect throughout the system and concentrate on delivering good providers for civil servants and the general public, significantly attending to grips with outdated, end-of-life methods.

“There’s a variety of unmaintained know-how the federal government depends on to do its job. It’s not ok in the meanwhile, and we clearly want to repair that. We have to try this similtaneously delivering worth to the general public,” he mentioned.

Whereas the federal government’s plans for digitisation focus closely on cyber safety and legacy methods, Pope mentioned it is very important suppose “a bit extra holistically” and concentrate on digital public infrastructure and what’s wanted to ship what works cross-government.  

The specialists additionally identified that whereas the remit of the brand new GDS is an efficient one, ought to it work, it is going to be troublesome to get proper.

Pope mentioned that traditionally, GDS has been targeted on citizen-facing providers, versus companies and the broader economic system, which has been ignored.

Gilbert added that the proof will likely be within the implementation.

“The opposite factor is, how are we monitoring whether or not it’s delivering on that remit in a means that doesn’t contain plenty of written paperwork and entails precise outcomes?” she mentioned.