Walk around the cavernous halls of the ExCel exhibition centre in London’s Docklands on any given weekday, and you are likely to encounter booths manned by eager sales people keen to share the benefits of information technology. DTX London, held in early October, is typical of these events, packed as it is with the lanyard-wearing masses.
Written on obligatory stress balls, canvas bags and notebooks – and on the banners and signs that adorn the booths themselves – you will find the language of IT. It often comes as single-word slogans – words like “simplify”, “modernise” and “unify”. Sometimes it’s more expansive – “maximise efficiency”, perhaps, or “minimise cost”. Maybe the two together. (DTX, incidentally, is a contraction of “digital transformation”, another go-to phrase.)
This is the world of IT and this is its language. It is as forceful as it is optimistic. The message is unequivocal: new technology will save you money and make you more productive. As we will discover, sometimes it does but often it doesn’t.
The language of IT has reached Westminster. Politicians cannot resist imitating some of its positivity and, even, indulging in some of its hyperbole. In his recent review of healthcare, Lord Darzi called for “a major tilt towards technology to unlock productivity”, while last month’s NHS consultation launch was framed by a promise to move the service from “analogue to digital”. Meanwhile, in his maiden address as Secretary of State for Science, Innovation and Technology, Peter Kyle said: “We need to rewire Whitehall, because technology is much more than just another sector to support or a strategic advantage to secure; it is the foundation for every one of our national missions.”
This is not new. A Cabinet Office paper published in November 2005 declared that 21st-century government is “enabled” by technology. The paper was commissioned by the then prime minister, Tony Blair, who more recently has been playing hype man to all things artificial intelligence (AI). Within 72 hours of Labour’s election victory in July he was encouraging readers of the Times into “the full embrace of the potential of technology”. His Tony Blair Institute for Global Change estimates that the UK stands to gain £40bn per year in public sector productivity improvements by embracing AI.
These productivity gains matter and not just for Whitehall and for town halls. They matter for the recipients of more effective services and for the Labour government’s wider growth agenda. A productive public sector can presage a more productive economy.
It is the promise of technology in this regard that proves irresistible. Last year Cat Little, then second permanent secretary at the Treasury, noted that public sector workers spend around eight hours a week performing administrative tasks. “Eight hours for core front-line public sector workers does not seem an effective use of their specialist skills and time,” she said.
There are instances when technology appears to be driving greater efficiency. Earlier this year, a government press release promised £1.8bn in benefits through a public sector productivity drive. This included saving 55,000 hours a year of administrative time in the justice system by digitising jury bundles, and a £100m dividend from using AI to reduce fraud. Meanwhile, in his ministerial speech, Kyle pointed to the impact of AI on Huddersfield Hospital’s radiology department where weekly scans have risen from 700 to 1,000 as a result.
Examples, large and small, point to the potential potency of technology. A recent civil service data challenge generated ideas including a proposal (project name: Posum) to use generative AI in order to summarise complex policies within the Department of Work and Pensions, so staff can respond more quickly to questions from citizens on how policy changes might impact their specific circumstances and individual benefit claims.
Despite these initiatives, public sector productivity is in a bad way. This in turn challenges the notion of technology as panacea. As of the end of last year, public sector output was 6.8 per cent lower than before the pandemic – and it remains unmoved from where the last Labour government left things nearly a decade and a half ago. It also compares unfavourably to productivity within the private sector.
This is a problem that needs fixing. The Office for Budget Responsibility estimates that by increasing public sector productivity by 5 per cent, the government could raise an additional £20bn in funding.
Technology, it is suggested time and time again, is the means of delivering the boost the UK requires.
Alongside the successes, however, there are plenty of failures. Take the attempts to modernise the NHS, for example. As we reflected in last month’s healthcare-focused issue of Spotlight, the National Audit Office has previously condemned the health service’s track record for digital transformation as “poor”, while an attempt to create a centralised patient record system was dubbed “the biggest IT failure ever seen”. It cost £10bn.
Other high profile failures include a shared services project that was designed to integrate human resources and financial services across the Department for Transport and its agencies. Nice idea. Unfortunately, projected savings of £57m turned into costs of £81m. The list goes on.
More than a generation ago, the American economist and Nobel laureate Robert Solow observed: ”We see computers everywhere except in the productivity statistics.” Solow’s remark caught the mood and was quoted in an 1993 MIT Sloan School of Management paper titled The Productivity Paradox of Information Technology. More than 30 years later, the paradox persists.
So why is technology still failing to deliver promised gains? Or, at least, failing to live up to the hype of its industry proponents? Broadly, there are reasons of implementation, which we will come to shortly. And there are reasons of definition.
In short, productivity in the public sector is different. “The great advantage of the private sector is that it has a very clearly defined bottom line. And that’s either the growth of the firm, profits the firm makes or whatever the shareholders decide the company should chase,” says Bart van Ark, managing director of the Productivity Institute, a UK-wide research organisation.
The public sector differs in two important ways. First, a department or local authority typically is a “multioutput organisation”, says Van Ark. In other words, it does more than one thing and delivers for all citizens not just a small subset of customers and clients. A police force, for example, is responsible for more than a dozen discrete functions, from traffic control to fighting fraud. It is ultimately for ministers to choose where to prioritise. “That’s a political decision you can’t put a price on,” notes Van Ark.
Second, the public sector has to think beyond inputs (labour, technology, equipment and so on) and outputs (surgery throughput, for example, or the number of police on the streets). It must think in terms of outcomes, too. It must deliver effectiveness, not just efficiency. Quality, not just quantity.
As the Productivity Institute puts it: “Public services need to deliver affordable, comprehensive, inclusive and high-quality services, often with an element of urgency and a recognition of rights.” Taken in this context, easy definitions, and measures, of productivity are difficult to come by.
Speak to technology specialists inside government, and they are likely to say the same thing. One IT director working for a prominent Whitehall department describes talk of productivity as a “category error”. Why? “Because we are not in the business of selling widgets,” he tells Spotlight. By speeding up response times, for example, a department may increase value to the citizen but it doesn’t necessarily reduce costs given more requests are likely to fill the vacuum created through efficiency. He describes this as delivering “more, differently, not fewer”, and that doesn’t save money.
Echoing the point, Van Ark offers the example of the Department of Work and Pensions. Over the recent past, it has introduced a great deal of new technology designed to process claims faster. Even when that proves successful, it is difficult to trace cause and effect – a causal link between more efficient claims-processing and speedier re-entry into the world of work, for example. Moreover, faster claims rates lead to greater asks from the public, says Van Ark, such as tailored services and faster response times, and also increased demand on the service overall.
We might call this the M25 problem – add another lane to an already busy motorway and more and more motorists will fill the newly available capacity.
As a result, the Whitehall IT director says, measuring productivity often involves a mix of extrapolation, assumption and implication. And this is not just a puzzle for central government. One senior technology leader for an inner London borough council, asks plaintively: “How do we improve residents’ outcomes using AI, and, ideally, how do we do it on the cheap?”
Problems of productivity are not just those of definition, however. There are failures of technology adoption, too.
Imogen Parker is the associate director for social policy at the Ada Lovelace Institute, independent researchers into the impact of data and AI. One of her areas of focus is automation. And like its less glamorous forerunners – such as robotic process automation (RPA) – AI has huge potential but inherent problems, too. There are two interlinked issues that particularly interest Parker.
First, she argues that inserting technology into any system changes the system. “Sometimes when people talk about productivity gains they imagine that you have static systems and you’re swapping like for like,“ she tells Spotlight. “The advent of email didn’t just mean that we had more productive letter-writing. It completely changed the way that we [communicate].”
The move from manual to automated processes creates ripple effects, she says. Unfortunately, these don’t result necessarily in higher productivity and lower costs. Instead, they change the nature of the processes. Users might defer to the tool, ignore it completely, or “game” it to get the desired results.
Second, she argues that automation removes something very valuable – “friction”. Ask an AI to draft a range of policy options, for example, and you are likely to miss a moment of inspiration that comes from hours of reading through the material. “Because I haven’t done the hard work to get there, it’s unlikely that I have internalised it,” Parker explains. “Sometimes you have to go through the pain to nail down what you think and identify what matters.”
With both issues in mind, she cites the Central Digital and Data Office – now part of the Department for Science, Innovation and Technology – which last year suggested that almost a third of civil service tasks could be automated. It is a punchy forecast but Parker believes it lacks rigour. Specifically, there is no real feasibility assessment or any attempt to put a cost on the transition. Measuring productivity gains without either offers ambiguity, at best.
Other problems of technology implementation persist. One is introducing new IT solutions without the necessary upskilling of staff or organisational adaptation. Another is what Matthew Taylor, CEO of the NHS Confederation, describes as a “false positives” problem where algorithms designed to isolate issues – and accelerate processes as a result – tend, instead, to over-index problem areas. “Technologies that we are promised are going to reduce demand often end up increasing demand,” Taylor told a New Statesman fringe event during the Labour Party conference in September.
So how do we fix it? How do we ensure that technology helps drive public sector productivity?
The first step is to avoid treating technology as a cure-all, or in isolation. Instead, a maxim that is well-worn in technology circles provides a useful starting point: people, process and technology.
Investing in people, and the skills they need, may prove politically difficult given current fiscal constraints, but it is likely to be key to unlocking long-term productivity gains. An estimated 5 per cent of civil servants work in digital and data compared to an economy-wide average of 8-12 per cent. “The public sector is largely a people sector; people deliver services,” notes the Productivity Institute’s Van Ark, “so you need to look at the skills of these people and their management.”
Management leads us to processes which need adjusting in order to get the most out of new technology. Sarah Woolnough, CEO at the King’s Fund, an independent health think tank, argues that it is essential to invest time in “not terribly glamourous change management” to benefit most from AI, for example. Changing the way an organisation operates may meet resistance, not least from leadership teams, but it is necessary to gain the confidence of a workforce that is likely to be wary of new technology – and to ensure that old processes don’t stymie new technologies. “Too often AI [adoption] is mimicking silos that exist in our healthcare systems,” she told the same New Statesman fringe event.
To this end, Van Ark urges simplification. “The simpler the processes are, the more likely you are to find success,” he says, nodding, as an example, to the Passport Office which turned a 360,000 Covid-era backlog into a five-day application process. “In a multi-outcome organisation we have to prioritise,” says Van Ark. “It’s a choice we’re making, one that political masters don’t always want to acknowledge.
Get people and processes right, and technology implementation is likely to be more successful. Or so the argument goes. Bart van Ark remains optimistic that technology can drive outcomes-based productivity but insists that there is no “silver bullet or quick fix… We shouldn’t over-hype this.”
“It would be a missed opportunity for the public sector not to use technologies that private sector organisations use,” he says. “If you want deliver services, you need to integrate the technology with the skills of your people… and the way your organisation is wired.”
Parker agrees. “You can’t just throw in the technology without thinking about what this means for the skills of the workforce, the way management operates, and the way the organisation operates,” she says.
On the use of AI, Parker says we should be both “curious and cautious”, adding: “We need to be much, much quicker about learning what works and what doesn’t work. The high-level rhetoric that some people are pushing almost paints this as a panacea that will increase productivity by a large number. I think that’s dangerous. I’m sure there will be occasions where AI can be beneficial, but let’s test it first.”
This article first appeared in our print Spotlight Party Politics Special, published on 15 November 2024.
Content from our partners
Related
