Responsible AI: Navigating the future of public services

As the government seeks to transform its operations with artificial intelligence (AI), public sector leaders are urgently developing frameworks to ensure AI is deployed ethically, transparently, and effectively.

How to achieve those goals was a topic under discussion at the recent Think AI for Government event, where experts gathered to address the responsible use of AI in public services (pictured).

“Responsible use of AI is a complex term,” said Tommaso Spinelli, senior AI change manager at the Cabinet Office. “It means making sure AI systems are safe, do not leak data, and meet legal expectations and requirements of government.”

The challenge goes beyond technical implementation. Sally Meecham, CTO at the UK Hydrographic Office, said that AI adoption requires a holistic organisational approach. “We need to create environments with clean data and lean into AI while proving its value,” she said.

Key to this approach is developing comprehensive guidelines that extend beyond technical teams. To the attend, the government has created 10 principles designed to guide civil servants across different expertise levels in responsible AI use.

Transparency is a critical concern. Steve Allison, global engagement manager at Adobe, highlighted three crucial areas: data use, model training, and output transparency.

“It’s not just about training large language models,” he said. “But understanding how to define response boundaries and ensure confidence in AI outputs.”

Adobe’s Content Authenticity Initiative represents one approach to addressing transparency. By implementing digital markers that allow users to verify AI-generated content’s origin, governments can combat potential misinformation risks.

But critically, responsible AI isn’t just a technological challenge – it’s a cultural transformation. “We need to take away the standalone technology perspective,” said Meecham. “Instead, we should focus on how AI creates efficiencies and supports organisational goals.”

This cultural shift requires breaking down traditional departmental silos. Spinelli highlighted the importance of cross-functional collaboration. “AI specialists need to understand policy challenges, and policy teams must comprehend AI capabilities,” he said.

To facilitate this, the government has launched an AI community bringing together professionals from diverse backgrounds. “We want people who aren’t necessarily AI specialists to participate,” said Spinelli. “We need different perspectives to make AI implementation right.”

Practical steps include developing AI ambassador programmes and embedding data scientists across different business units. However, ethical considerations extend beyond immediate operational concerns – organisations must also consider environmental impacts, data privacy, and potential workforce transformations, said the panel.

Government estimates suggest digital transformation and AI could generate £45 billion in savings. But realising these benefits requires a nuanced, responsible approach. Key recommendations emerging from the discussion include:

1. Develop clear, accessible guidelines for AI use across all organisational levels
2. Prioritise transparency in AI implementation
3. Create cross-functional teams to support AI integration
4. Invest in training and change management
5. Continuously assess ethical implications and potential risks

Allison warned against a one-size-fits-all approach, noting the need to distinguish between different AI technologies. “Be clear about what you’re trying to achieve,” he said. “Different AI technologies—machine learning, generative AI, and autonomous systems—require different considerations.”

Added Spinelli: “Getting AI right means getting the right tools but also implementing the right change management strategies.”