As AI moves from experimentation to implementation across public and private sectors alike, expectations continue to mount across Whitehall that the technology can (and indeed must) be harnessed to drive the scale of service transformation needed to deliver on the government’s mission-based modernisation agenda, especially in financially straitened times.
However, can departments realise AI's potential for smarter, more proactive services whilst navigating the myriad challenges that still abound across the sector (legacy systems, data fragmentation, and fragile public trust in big data initiatives key amongst them)?
There remains much non-trivial (and unfashionable) spade work to be done across the sector to ensure that data quality, infrastructure, and governance are up to scratch before any production-level AI can add significant value. Unfortunately TechMarketView estimates that the public sector ‘Solutions’ market will achieve a CAGR of only 1.7% from 2024-28 (representing a 0.4% contraction in real terms) – and, with these initiatives focusing on many of the cloud migration and associated digital transformation projects that set the data foundations for AI initiatives – there’s a risk that promising projects will fail to find their footing when they attempt to transition from pilot to production.
So, whilst there are breakthrough opportunities to understand, and early successes to learn from and replicate, significant barriers still remain. Technology leaders across central government and beyond, therefore, need to understand how, where, and why AI will work for them (as well as how, where, and why it won’t).
Craig Wentworth, Research Director at TechMarketView
Moving upstream
The promise of AI in government extends far beyond efficiency gains in business-as-usual processes. It represents a fundamental shift in how government identifies, prioritises, and addresses citizen needs before crisis points (where remedial measures ramp up costs).
An AI-powered shift from reactive to predictive service delivery can reduce demand for expensive downstream interventions by targeting at-risk individuals and recommending early preventative measures. This ‘prediction and prevention paradigm’ sets the context for much of the underpinning role new technology is expected to play in support of the government’s 10-Year Health Plan, for instance – with scope for transferability far beyond the NHS (in terms of tech and techniques for fraud detection, social care in local government, etc.). Yet scaling these approaches requires confronting uncomfortable truths about current capabilities.
Battling the current
However, beneath promising pilots lies an uncomfortable reality. The State of Digital Government Review reveals that 70% of public sector organisations operate with what they consider to be fragmented, uncoordinated data landscapes. These ‘Cinderella Systems’ – unloved, outdated, yet still at the heart of many unmodernised government services and responsible for processing millions of critical transactions – represent decades of technical debt that AI cannot simply bypass. Without addressing these foundational weaknesses, AI risks becoming expensive (and potentially unreliable) window dressing on crumbling infrastructure.
Also, with cybercriminals exploiting new tools with fewer guardrails, AI in cybersecurity presents threats as well as opportunities. Public sector organisations – like their commercial counterparts – must simultaneously deploy AI for cyber defence (in order to cope with ever-increasing attack surfaces), whilst also protecting against AI-enabled attacks themselves – a challenge which requires new thinking about security architecture and risk management.
AI’s pre-deployment considerations aren’t confined to the technical; the human dimension proves equally complex. Whilst some Civil Service trials (focusing on GenAI’s more mature use cases with cross-sector applicability – such as summarising meetings, drafting documents, etc.) show early promise, with significant time savings reported; broader challenges loom. Many UK workers have never used generative AI, and there are significant demographic disparities in adoption, meaning that skills gaps need addressing lest a new AI divide be created (exacerbating the earlier digital one). There are deeper questions about widespread AI use too. Some end users praise AI tools’ apparent ability to remove human bias from decision-making… but can we / should we ‘trust the machine’ (yet), and what does this mean for democratic accountability?
The right stuff
Despite these challenges, there are early implementations that do demonstrate AI's transformative potential when deployed thoughtfully; some even representing fundamental re-imaginings of operational processes and service delivery, and an attendant shift in organisational culture. The AI opportunity extends beyond individual departmental wins, too. With proper data foundations in place, AI has the potential to enable government to work as a connected system rather than isolated silos.
The key is in selecting use cases that deliver measurable value for early milestone wins, whilst building the trust and capability needed for more ambitious applications that build off initial transformations. The evolution from simple automation to genuine public service augmentation and enhancement follows a similar pattern to how, with cloud migrations, initial workload efficiencies can often then lead to later-stage benefits from wholesale adoption of a wide range of cloud services (which simply weren’t on the table when everything ran on-prem).
The path forward requires an honest reckoning with these challenges, a full understanding of the applicability and transferability of successes, and tech partners prepared to provide realistic appraisals of the scale and scope of the work to be done before real benefits can be seen (and quantified). As citizen expectations rise and fiscal pressures mount, the question isn't whether government should embrace AI, but whether it can do so in the right places, in the right way – and quickly enough – whilst maintaining the trust and accountability that society demands.
AI may not necessarily be your saviour; but nor need it be your scourge.
Come and find out more at the Future Gov Summit. Register for free HERE.
You’ll also hear real-life examples from departments like TfL, MOD, and GIAA, who are already delivering AI successfully - and get the latest research and insights from TechMarketView, Microsoft, AWS, ServiceNow and Salesforce on the AI opportunity for government.