Starmer to announce AI should replace some work of civil servants

The UK government’s latest push for AI-driven efficiency in Whitehall is yet another example of ambitious rhetoric without a clear strategy for execution.

Prime Minister Keir Starmer is set to announce that AI should replace some of the work of civil servants so that it can be done better, quicker, and to the same high quality and standard. While politically appealing, this language risks oversimplifying the role of AI in public service and overlooking the essential role of human oversight, investment in AI skills, and the UK’s own AI ecosystem.

AI is not a magic solution for reducing government inefficiencies. Instead, it is a tool that, when used appropriately, can enhance productivity and improve services. However, it still depends on human expertise and ethical oversight. The key discussion should not centre on whether AI will replace civil servants; instead, it should focus on how AI can support them by allowing them to concentrate on high-value tasks while automation takes care of repetitive processes.

AI is a Tool, Not a Replacement

AI will require humans to be in the loop in the foreseeable future. The idea that AI can “replace” civil servants fails to acknowledge the limitations of even the most advanced systems.

• Decision-making complexity: AI lacks full context and nuance, particularly in high-stakes policy and regulatory decisions.
• Ethical considerations: AI systems must be carefully monitored to avoid bias and unintended consequences.
• Validation and feedback: Human oversight is necessary to review AI outputs, ensuring accuracy and fairness.
• High-stakes scenarios: Government decisions—such as benefit entitlements, social care assessments, and law enforcement applications—demand human judgment that AI alone cannot provide.

While AI can undoubtedly increase efficiency in back-office tasks, automate routine processes, and assist with data analysis, it cannot replace human expertise in policymaking, case management, or service delivery.

The Right AI Strategy: Investment in People, Not Just Technology

If the UK government is serious about leveraging AI to improve the civil service, then investment must go beyond procurement and focus on building an AI-ready workforce.

The announcement includes 2,000 new tech apprentices, which is a step in the right direction but far from enough to ensure a long-term pipeline of AI expertise in government. The UK should create AI talent pathways in universities, apprenticeships, and public sector training programmes.

The UK must also invest in its private sector AI talent and infrastructure. Instead of handing AI contracts to large foreign firms, the government should invest in UK-based AI innovation and minimal and medium-sized enterprises (SMEs) that can deliver domain-specific solutions tailored to government needs.

AI should be a strategic enabler deployed where it delivers the most impact, such as streamlining government data systems, improving decision-making with analytics, and enhancing citizen services. AI adoption must be targeted and problem-led, not a broad brushstroke applied indiscriminately.

The Risk of Overpromising and Under-Delivering

One of the most significant risks with Starmer’s approach is that it oversells what AI can achieve, leading to unrealistic expectations and public disappointment. We’ve seen this before—large-scale government technology projects often fail due to poor planning, lack of expertise, and insufficient focus on real-world implementation.

A more effective approach would be:

Take an iterative approach – AI adoption should be incremental, tested through pilot projects, and refined based on measurable outcomes rather than rolled out wholesale across the civil service.

Identify where AI will have the most significant impact – Focus on improving data-driven decision-making, reducing inefficiencies in administration, and enabling frontline workers to do their jobs more effectively.

Develop clear AI governance frameworks. Ensuring ethical, explainable, and responsible AI adoption within government is essential to maintaining public trust.

Invest in skills development and AI SMEs – The UK has an emerging AI startup ecosystem. Why not build government AI capability in-house rather than relying on external tech giants?

AI Cost-Savings: A By-Product of Efficiency and Productivity

The UK government should first focus on transformation and productivity rather than positioning AI as a pure cost-cutting tool. Well-implemented AI and data strategies will naturally lead to efficiency gains and cost savings, but making financial reduction the primary goal risks undermining long-term service quality.

A modern data infrastructure, combined with AI-driven decision-making, can streamline processes, reduce duplication, and improve citizen outcomes. However, AI is implemented with the primary intent of cutting budgets. In that case, it may lead to short-sighted decisions that result in more inefficient services, reduced capacity, and poor value for taxpayers. The focus must be on building better, more responsive public services, with financial efficiencies as a natural consequence of good design, not an artificial target.

AI Adoption Should Be Thoughtful, Not Reactionary

AI has enormous potential to improve government services, but it must be deployed strategically and ethically, and its limitations must be clearly understood. Sweeping AI-first policies without careful planning risks failing both civil servants and the public they serve.

If the government truly wants to embrace AI responsibly, it must:

• Invest in UK-based AI skills and innovation.
• Focus on augmenting, not replacing, civil servants.
• Ensure AI adoption is outcome-driven, not technology-driven.
• Maintain transparency, ethics, and accountability in AI decision-making.

At Modular Data, AI should be a force for good in government and implemented to benefit society. AI should serve as an enabler of public sector transformation, not a justification for job cuts.

The conversation must move beyond political soundbites and focus on practical, responsible, and citizen-focused AI deployment.