Overall, perceptions of AI were favourable. Among those who had encountered AI at work, 58% said their experience had been positive, while just 14% said it has been negative. There was also optimism about the potential future impact of AI. Nearly two in three (62%) agreed that AI will improve how their department serves the public, while just 17% disagreed.

FDA General Secretary Dave Penman comments that the research “shows that FDA members are not blockers to progress – indeed, there is a strong appetite amongst civil servants to utilise AI to deliver better services to the public”.

However, he also reflects that “rollout is inconsistent across departments, which limits productivity gains, gives little incentive for innovation, and fails to address the very real concerns people have about adopting the new technology”.

A theme that emerges in the report is that change is being done to workers, not with them. Just 29% of those surveyed had been consulted on artificial intelligence tools in the workplace. Some also felt that engagement, where it occurred, was not sufficiently meaningful. One survey respondent reported: “I have attended many meetings with my line managers [and] it is clear that the decision has already been made [and] the meeting is simply a way to pay lip service to show they have ‘engaged’.”

This is a commonly held frustration, as 66% said they wanted to be more involved in shaping how their organisation adopts AI.

The report also finds that the government’s vision is not widely understood. Indeed, one respondent reported: “A leadership tone of ‘Let’s use AI’ is positive but unfocused, and I do not understand what my department wants to see AI being used for, and what outcomes it expects.”

Access to personal AI tools is characterised in the report as “frustratingly ad-hoc”, with current processes for securing funding and permission to use new technologies creating a fragmented landscape with significant variation in experience.

33% of respondents who had not used AI to improve how their team works said this was because they did not have access to the right technology.

One respondent reported: “[It] is not accepted in my department, but has been licensed for use in other departments who clearly recognise its benefits.” Another said: “The pace and scale of rollout doesn’t match the comms. It privileges policy teams.” A third expressed concern that “only senior civil servants (DDs and above) have access” to certain tools.

The research also captured concerns within the civil service for what the adoption of AI could mean in practice. 9% of respondents say they are at least “slightly” concerned about AI use in their organisation. Nearly half (49%) describe themselves as “moderately”, “very” or “extremely” concerned (26%, 11% and 12% respectively).

Among those who have not yet used AI to improve how their team works, 36% say this is because they are worried about risks to the public. However, only 3% of respondents provided examples of things that had actually gone wrong, while 56% gave hypothetical examples – a pattern the report says underlines the case for stronger engagement, clearer safeguards, and better communication.

The Fabian Society argues that these concerns reflect “insufficient clarity, consultation and communication”, rather than widespread evidence of harm – and that a serious reform programme needs staff voice built in from the start.

Adopt, Innovate, Transform puts forward a set of recommendations to build a genuine ‘move fast and fix things’ culture.

It calls for a ‘digital dividend’ for delivery, which would mean civil servants receive performance-related bonuses if they meet agreed delivery milestones. The report argues this would strengthen incentives to stay in post and complete transformation initiatives, with milestones set at recruitment or at project commencement.

As another way to reward delivery, the report also argues that departments should be able to keep AI savings through a digital transformation reinvestment framework. This would allow departments to retain and reinvest efficiencies generated by AI – as is commonplace in business – rather than automatically returning savings.

Another central challenge the report seeks to address is how to up-skill the workforce and cultivate talent in priority areas. It recommends the government should identify key skills relevant to specific professional standards and develop modular courses for staff wishing to develop these skills at different levels. It should allow staff to access courses relevant to their role for free via the National School for Government and Public Services, and guarantee everyone can spend an appropriate amount of time for their role on these courses. It proposes the government considers at a baseline least of 2.5% of working hours, with a higher allocation for those in priority areas.

Cognisant of the very real concerns civil servants have about the adoption of AI in government,  Adopt, Innovate, Transform also suggests ways to improve transparency by working with trade unions to agree and codify appropriate use for AI in HR and establish formal mechanisms to track the impact of AI on workers – including diversity impacts and potential for racial bias on such lines – and keep the need for additional support under review.

Of course, all of these recommendations will only make a difference if staff have access to the correct tools. The reports calls on the government to prioritise the development of internal solutions assessed to have the greatest potential productivity impact, and concurrently scope trials of external solutions to test impact across a range of professions, grades and departments. This should be guided by clear and deliberate consideration of accountability and UK government sovereignty.

Once tools are proven safe and effective, a key recommendation is to make them available across government through a centralised repository of approved tools, funded through a cross-government pot, with continued access contingent on light-touch reporting to monitor impact and manage risk.

As Penman argues, “the government’s stated desire to ‘move fast and fix things’ can only happen if civil servants are given proper access to the tools and training they need to truly deliver a smarter state”.