The transformative power of generative AI in the public sector

With the potential of generative AI capturing the imagination of business, James W Hewitt, associate partner at Go Reply, explores the benefits to the public sector.

Posted 31 January 2024 by Christine Horton

Generative AI has captured the public’s imagination since generally available Large Language Models base on Google’s Transformer such as ChatGPT and Bard became advanced enough to write convincing answers to conversational questions. Despite widespread public adoption, the public sector is commonly perceived as lagging behind the private sector in integrating these tools into its systems and processes. This is about to change. The government, the civil service, and the wider melange of public sector agencies in the UK stand at a turning point, where the state is beginning to realise the benefits that generative AI can offer, resulting in greater efficiency and better public services.

Public sector applications of generative AI

Generative AI can redefine the way public services operate. By modernising manual processes, the technologies can pave the way for a step-change in service delivery.

For example, consider the fact that competing departmental demands mean that both mandarins and ministers are increasingly pressed for time. This can result in analysis that is less thorough than otherwise would be the case in a stable economic and geopolitical environment, resulting in less-than-optimal outcomes for public services. Generative AI can be used to summarise vast datasets, whitepapers, survey responses and other content to offer a clearer and quicker understanding of complex topics. By analysing data, generative AI can also unearth insights that might otherwise be overlooked, enhancing the capacity for cost-benefit analysis of new policy proposals. Combined, the suitability of generative AI-enabled summary and data analysis can enable more considered, yet swifter decision-making in areas that until now may have fallen by the wayside.

Through a wider lens, generative AI can help public sector teams break free of bottlenecks automating certain business processes. With repetitive, non-cognitively challenging administrative tasks delegated to the technology, civil servants of all levels can focus on delivering more strategic and innovative initiatives, resulting in more value for the citizenry.

Generative AI can also directly benefit those outside government. For example, advances in natural language processing means that citizens can access public services more swiftly; requests from multiple channels can be answered directly by the technology or expedited to the appropriate handler. This ensures citizens receive timely and accurate support for their specific problem, boosting satisfaction and ultimately their trust in government.

Key safety considerations

When we consider how to maintain this trust, it’s important to note that the public sector’s cautiousness towards technology stems from a responsibility towards its citizens that is distinct from the business-customer relationship. While generative AI offers numerous advantages to the current status quo, its implementation must be approached with care. For example, LLMs can ‘hallucinate’ and return inaccurate information. It is vital that guardrails are implemented and systems are thoroughly tested and monitored to avoid poor outcomes.

Moreover, intellectual property concerns, especially when AI systems learn from vast datasets, require careful navigation to avoid copyright and ethical issues. It is also imperative that government develops skills programmes in concert with technology providers, so that this step change in capability remains understandable and accessible to every citizen, ensuring no one is left behind in this technological evolution.

Prospects for successful implementation

For effective generative AI implementation, the public sector must equip its workforce with the necessary skills to maximise the potential of the technology. The emerging field of prompt engineering, which involves crafting effective input prompts for desired chatbot responses, is one such area that will become foundational to success.

Also foundational will be the adoption of Large Language Model Operations (LLMOps). Analogous to DevOps, organisations must develop this capability to curate, maintain, provide guardrails and operate a technology with new requirements and pitfalls.

Smart regulation will play a pivotal role in striking the balance between unprecedented innovation and the imperatives of change management. Ethical application, transparency, accountability, and bias avoidance will be at the forefront of these discussions, ensuring that AI not only benefits citizens but also respects the institutional norms and values the government embodies.

Amidst an uncertain global economic outlook, the potential to improve the public sector through integrating generative AI is clear. It’s essential that this process is handled with both the enthusiasm warranted, but also caution to the evident risks and uncertainties. By taking a mindful approach, the public sector can harness the full power of generative AI, ensuring an efficient, useful, and more effective government for the future.