Healey’s AI bet won’t pay off without the right data

Share this article

John Healey’s Budget repeated the message that the Government sees AI as a route to economic growth and better public services. From investment in British AI capability to plans for data centres and AI adoption across Whitehall, ministers are putting money and political weight behind that ambition. The success of this investment will depend on work that rarely makes headlines – ensuring the data behind this technology is accurate, accessible and properly maintained. Without sustained attention to these foundations, the Government risks spending heavily on technology that cannot deliver the improvements it promises.

At the Open Data Institute (ODI), we have long argued that data needs to be treated as infrastructure because public services, businesses and communities depend on it to make decisions and deliver essential services. Like roads and electricity networks, it needs ongoing investment and maintenance to remain useful. Government departments and public bodies need named responsibility for the data they collect and publish, including keeping it accurate, documenting its limitations and making it accessible to those entitled to use it. The Government’s AI programme depends on this work, yet the people doing it rarely feature in ministerial announcements.

The Budget’s commitment to requiring councils to publish key planning documents as open data by the end of 2027 illustrates the opportunity available to the UK of using its data more effectively. Alongside this, ministers intend to make the PlanAssist AI tool available to all English councils by the same deadline, subject to further testing. Done well, opening up planning data could help residents understand development proposals, enable smaller businesses to build useful services and give researchers a clearer picture of where the planning system is struggling. Its value could extend well beyond the AI tools ministers want councils to adopt, but a publication deadline will achieve little if neighbouring councils describe the same things differently, records cannot be compared or nobody has the resources to keep them current.

Delivering this commitment will require common standards, clear licences and staff with the time and skills to maintain the information. Councils also need clarity about what they must publish, how quality will be assessed and who will support them. Central government could fund shared tools and technical support, reducing the need for each council to solve the same problems independently, alongside targeted funding where local capacity is weakest. The cost of preparing and maintaining data needs to be included in the implementation budget from the outset, rather than becoming another demand on stretched council resources.

The same question arises across Whitehall. The £35 million to establish GovAI and £50m Civil Service AI transformation fund signal a determination to change how government works, alongside plans to save a further £600m from back-office budgets. Those programmes offer an opportunity to fund the data preparation and maintenance on which successful adoption depends. Assessing these costs before committing to deployment would give ministers a more credible basis for estimating savings. Each substantial public-sector AI programme needs to establish whether its data is fit for purpose, who is responsible for it and how its upkeep will be funded.

The same scrutiny applies to sovereignty. The Budget’s backing for British AI companies is welcome, and public investment can also secure lasting control over the systems on which public services depend. Keeping sensitive data in the UK matters, but the location of a server only tells us part of what we need to know. Public bodies also need to understand who controls access, whether a system can be independently examined and whether they can change suppliers without losing access to essential information or facing prohibitive costs. Procurement can secure those freedoms, through contracts that guarantee appropriate scrutiny, usable routes for transferring data and credible plans for changing suppliers. Spent this way, public money strengthens the UK’s capacity to make choices about technology over time.

The public benefit of AI infrastructure deserves equal attention. The Budget’s new Data Centre Covenant offers enhanced government support to strategic projects in return for commitments on community benefit, environmental standards and other national priorities. That bargain needs to be measurable and enforceable. Communities being asked to accommodate data centres will want to assess their effects on local energy and water resources, alongside promises of jobs and investment. That depends on environmental reporting that is consistent enough to compare facilities and useful enough to hold operators to account.

As more services adopt AI, people’s ability to get help matters as much as the technology. The Budget’s local AI partnerships, beginning in Yorkshire, offer an opportunity to involve communities in shaping adoption and identifying where support is needed. Part of their success will lie in determining who benefits and who remains excluded. Someone who lacks digital skills, struggles with English or cannot afford reliable connectivity should still be able to understand a decision and challenge it. Digital inclusion requires sustained support and accessible human assistance, especially when something has gone wrong.

None of this diminishes the opportunity. The new £15m AI Challenge Prize, beginning with the NHS, could help focus innovation on practical improvements to public services. Better data could help clinicians assess treatments, councils plan homes and businesses develop services that save people time and money. Reliable AI could make expertise more widely available and help public servants devote more attention to the people who need them.

The Budget provides investment and political momentum. The Government must now give the less glamorous work of maintaining data, setting standards and securing public value the same attention as the technology itself. Creating a Minister for Data would be a smart place to start.

The test for the Government will be whether services become more accurate, accessible and accountable, and whether the benefits reach people who currently struggle to get support.

That is how the Government should measure the return on its AI investment.

Share this article