A major government trial has demonstrated that AI coding assistants are significantly enhancing productivity among developers across the public sector, saving nearly an hour of work per day per coder, which is equivalent to 28 working days each year overall
The trial, conducted by the Department for Science, Innovation and Technology between November 2024 and February 2025, involved over 1,000 software developers and engineers from 50 different government departments.
Participants used AI tools, including GitHub Copilot, Microsoft’s AI assistant, and Google’s Gemini Code Assist, to support their daily coding tasks.
Building technology faster and smarter
The increased efficiency allowed developers to focus more time on building critical new technologies for public services.
This included tools such as the Whitehall’s Humphrey AI assistance, healthcare systems, and digital infrastructure improvements. These projects are all part of the government’s “Plan for Change” initiative, which aims to modernise the public sector through more innovative use of technology.
AI assistance primarily assisted developers by drafting initial code versions and reviewing existing code for improvements or errors. While only 15% of AI-generated code was used without any changes, this demonstrates that human developers were actively reviewing, editing, and validating code to maintain high standards of accuracy and reliability.
Positive response from public sector developers
Feedback from participants was primarily positive. Nearly three-quarters of developers found the tools offered good value for their organisation, and over half said they would prefer not to return to coding without AI assistance. Many reported that tasks were completed quickly and problems were solved faster because of AI support.
In total, over 1,250 AI assistant licenses were used during the trial period, with the large majority being GitHub Copilot, followed by Gemini Code Assist. These tools offered developers a way to reduce repetitive tasks, improve code quality, and focus on more complex aspects of their work.
Technology Minister Kanishka Narayan said:
“For too long, essential public services have been slow to use new technology – we have a lot of catching up to do. These results show that our engineers are hungry to use AI to get that work done more quickly, and know how to use it safely.
“This is exactly how I want us to use AI and other technology to make sure we are delivering the standard of public services people expect – both in terms of accuracy and efficiency”
A more efficient public sector
The findings support the UK government’s ambition to use AI across public services as part of its long-term artificial intelligence strategy. With the public sector aiming for up to £45 billion in efficiency savings through the adoption of more innovative technology, AI coding tools are proving to be a key part of that vision.
The productivity improvements also contribute to reducing backlogs, accelerating innovation, and delivering higher-quality services to citizens. The trial revealed that developers were not merely using AI tools unthinkingly; they received dedicated training and support to utilise them responsibly and effectively. Government guidance made clear that developers remain fully accountable for the code they deliver, even when assisted by AI.
The success of the AI coding assistant trial is expected to shape future government decisions on wider AI adoption. The trial was the largest of its kind for Google’s Gemini Code Assist in the UK public sector, highlighting the value of strategic partnerships between the government and leading technology providers.
