Artificial intelligence is already reshaping public services, from streamlining planning permissions to accelerating NHS diagnostics. But to fully realise the potential of AI, the government must address the digital skills gap that persists across the public sector through effective training.
Lewis Hodges, senior consultant at AtkinsRéalis
A House of Commons report in March 2025 on the ‘Use of AI in Government’ highlighted the skills gap as a key blocker to successful AI adoption. The report revealed that 70% of departments struggle to recruit and retain staff with AI expertise, and that in 2024 alone, half of the roles advertised in civil service digital and data campaigns remained unfilled. These statistics signal a growing gap between ambition and capability.
Through initiatives spanning industry, research, and the public sector, the Department for Science, Innovation and Technology (DSIT) has already taken significant steps in building an AI-aware civil service. In June, Innovate UK, aligned with DSIT’s national strategy, launched the AI Skills Hub. The Cabinet Office has also announced AI as the focus of this year’s ‘One Big Thing’ project. Rolling out in autumn, the objective is to provide all civil servants with a ‘working knowledge’ of AI.
These initiatives mark an important step in building foundational AI awareness across the civil service. However, they primarily focus on informing knowledge rather than embedding behaviours.
To truly prepare civil servants for the practical and ethical challenges of AI, the next step is to transform theory into applied skills and habits. Although there is no silver bullet, tailored, scenario-based learning approaches excel in bridging the gap between conceptual understanding and real-world decision-making.
Hands-on learning for MoD
Scenario-based learning changes behaviours by immersing participants in situations that mirror their daily challenges. Unlike traditional methods such as lectures, static e-learning, or playbooks, which encourage passive memorisation, scenario-based learning demands active decision-making and problem-solving.
Reports from Google and The Alan Turing Institute stress that AI upskilling must be interactive, role-specific, and behaviorally anchored. This ensures learning is immediately applicable, driving real performance improvements in the workplace.
Pilot rollouts of AI can have mixed results. Without the right training, staff may struggle to unlock the full value of tools during experimentation phases, leading to inconclusive results even when genuine efficiency gains are possible. Embedding practical training early helps ensure that pilots reflect actual tool capability, not just user readiness.
A standout example of scenario-based learning is the Ministry of Defence’s (MoD) Cyber AB&C programme, which was delivered in partnership with AtkinsRéalis. The most recent iteration was completed in July 2025.
Designed to shift behaviours around cybersecurity, a gamified escape room format immersed personnel in active problem-solving based on real-world scenarios. The scenarios were co-designed with MoD teams to reflect the specific operational contexts and behavioural risks faced by personnel, and were offered to participants as selectable modules, enabling them to tailor their learning to areas of greatest relevance.
Academic research consistently shows scenario-based learning drives higher participation, often exceeding 90% engagement compared to much lower rates in e-learning and lecture-styled formats. UCL notes that in scenario-based learning, participants “find things out for themselves” and that it fosters a deeper understanding by letting learners explore the consequences of their choices.
Despite its proven benefits, scenario-based learning remains underused in government due to perceived complexity and resource constraints. However, in the case of MoD, the Cyber AB&C programme saved around £500,000 by consolidating previously disparate cyber advice contracts. The programme was recognised with the Management Consultancies Association Award for Change and Transformation within the Public Sector 2022.
Defining AI fluency
AI fluency is more than just understanding what AI is. It’s about being able to confidently and responsibly apply AI tools, principles, and thinking in day-to-day contexts. In the public sector, this means using AI not just as a technical solution, but as a strategic enabler of better policy, service delivery, and decision-making.
An AI fluent government would exhibit:
- Understanding: Comprehension of what AI is, how it works, and where it can (and can’t) add value. This includes foundational knowledge of AI ethics, risks, and opportunities.
- Confidence: Embedding AI into everyday workflows, decisions, and problem-solving. This requires not just skills, but motivation, practice, and the right organisational conditions.
- Culture: AI is trusted by both staff and the public, and is applied cross-functionally without gatekeeping. There are clear standards, collaborative norms and inclusive governance measures.
Scenario-based learning is a powerful enabler of this vision. By embedding AI concepts within real-world public sector challenges, it turns abstract theory into practical, role-relevant skills. It builds confidence through hands-on experience and fosters a culture where AI is understood and applied as a shared capability, not confined to technical specialists.
Over time, this approach can help to close the digital skills gap by equipping a broader range of civil servants with adaptable competencies and the confidence to engage with AI – reducing reliance on external expertise and enabling government to cultivate its own resilient, future-ready AI workforce.