National Audit Office AI report: Applying government design principles to public services

The National Audit Office (NAO) recently published a report on artificial intelligence (AI) in government (See video summary below). This report offers valuable insights into the current state of AI adoption across UK public services. As someone who has worked extensively with government digital standards, including implementing GDS-inspired service standards in the Royal Navy (you can read more about that project here), I'm particularly interested in exploring how we can apply Government Design Principles to make AI work better for users and service teams. I've gone through some of the principles in relation to what the report is saying.

Start with user needs

The NAO report reveals that only 37% of surveyed government bodies have deployed AI. This low adoption rate presents an opportunity to get things right from the start:

• AI should solve real user problems, not be used for its own sake.

• Begin by understanding user needs and pain points.

• Choose AI only when it’s the best solution to meet those needs.

Design for inclusivity

The report highlights concerns about potential bias in AI systems. To address this:

• Involve diverse user groups in research and testing.

• Consider the needs of users with low digital skills or limited access.

• Design AI solutions to work for everyone, not just the majority.

Make things open

43% of surveyed bodies identified lack of public trust as a barrier to AI adoption. To build trust:

• Be transparent about where and how AI is used.

• Explain decisions in plain English, not technical jargon.

• Provide clear routes for users to question or challenge AI-driven decisions.

Do the hard work to make it simple

The report identifies legacy IT and data quality issues as significant barriers. Before implementing AI:

• Address underlying data quality and access problems.

• Modernise legacy systems where possible.

• Ensure the digital infrastructure can support AI initiatives.

Iterate and improve

While only 37% of bodies have deployed AI, 70% are piloting or planning AI initiatives. This is a positive sign, but it’s important to:

• Begin with small, focused pilots.

• Set clear success criteria linked to user outcomes.

• Iterate based on real-world feedback and data.

Understand context

The NAO found that many organisations lack specific AI governance structures. To ensure responsible AI use:

• Establish clear accountability for AI decisions.

• Develop ethical guidelines tailored to organisational context.

• Regularly review and update approaches as AI capabilities evolve.

Data drives decisions

The report suggests a lack of clear performance metrics for AI initiatives. When implementing AI:

• Focus on outcomes for users, not just operational efficiency.

• Measure the impact on service quality and user satisfaction.

• Be prepared to change course if the data shows AI isn’t delivering value.

By following these principles, we can work towards AI implementation that truly serves users and delivers on the promise of improved public services. As the NAO report shows, we’re still in the early stages of this journey. That means there’s a golden opportunity to get the foundations right and build AI-enabled services that are efficient, ethical, and user-centered.

The goal isn’t to use AI for its own sake, but to create better services that meet user needs. Keeping this focus is crucial as the public sector navigates the AI frontier.

Report NAO Director, Ruth Kelly, outlines our findings on the use of AI in government.