The UK’s new Labour government is warming up to the ‘white heat’ of technology. During the election, Labour leaders talked about using AI to tackle truancy, help jobseekers, and analyse hospital scans. Peter Kyle, the new Secretary for Science, Innovation, and Technology, praised technology’s potential to save time and make interacting with public services a ‘more satisfying experience’.
This cautious optimism is a good sign, especially compared to the bold claims about AI’s ‘truly revolutionary’ potential. The new government has a chance to shift public sector AI policy towards longer-term and more strategic goals. However, to seize this opportunity, they need a clear understanding of the current situation and a realistic assessment of how these technologies work when used in public services.
Current use of AI in the public sector
The Ada Lovelace Institute has gathered considerable evidence over the past five years on how AI is being used in the public sector. This includes everything from NHS data analytics to AI-powered health predictions, COVID-19 technologies, and live facial recognition. They plan to release a review of their findings later this year, which will offer lessons for the new government.
To start, the new administration should take a look at the current state of AI in public services. The National Audit Office (NAO) has reported significant waste in the UK due to poor infrastructure and mismanaged projects. Local councils face financial pressure and strikes in key services are common due to low pay and poor conditions. Issues like an ageing population, economic challenges, COVID-19 aftermath, and past austerity measures are all putting a strain on public services.
The appeal and challenges of AI
Given these challenges, the idea of an AI revolution to improve services and efficiency is appealing. The new government faces tight budgets and resource constraints both external and self-imposed, making it hard to increase spending on public services. If – and it’s a big if – AI can help solve these problems, it would be wise to explore its potential.
So, how AI could help the public sector escape the ‘doom loop’ of underfunding and struggling services? By improving efficiency or productivity through automation, like supporting civil servants with document analysis to save time and money. It could also enhance the quality of a service by enabling new innovations, such as better cancer diagnostics or personalised job-seeking assistance via chatbots.
Key questions and concerns
However, it’s important to question whether these AI applications actually work in practice. Do they work well enough for everyone, and do they perform effectively in context – in real-world settings like hospitals and schools? These questions are crucial for AI used in the public sector, where decisions can deeply impact people’s lives.
Despite growing interest, there’s little empirical evidence on AI’s use in public services. A survey by the NAO found that over a third of public bodies were using AI, with another third piloting or planning to use it. However, there’s no comprehensive resource tracking where and how AI is being used, even within central government. This lack of oversight makes it difficult to evaluate the effectiveness of AI tools.
The need for better monitoring
The new government should prioritise understanding how AI is currently used in public services. Coordinating digital, data, and AI efforts under one roof in the Department for Science, Innovation, and Technology (DSIT) can help. Implementing the Algorithmic Transparency Recording Standard (ATRS) across the public sector is also essential. This involves documenting how algorithms work, their decision-making processes, and the data they use, ensuring accountability and accuracy.
Delivery teams should complete the recording standard and keep records updated as AI tools evolve. This will support better monitoring and evaluation. Additionally, the DSIT should conduct a review of AI in government and continue to oversee its deployment.
Addressing market imbalances
The opaque market for AI products complicates things further. Public sector decision-makers often rely on sales pitches from private firms, lacking detailed information about AI systems. Regulation of AI firms is necessary to ensure these systems are safe and effective before they reach the market.
Binding regulations on AI developers, as promised in the 2024 King’s Speech, are a crucial step. Establishing a statutory footing for the AI Safety Institute (AISI) with mandatory access agreements for AI developers can support this. The government should also set strict conditions for AI suppliers, demand transparency, and invest in reducing the skills gap between the public and private sectors.
Building a new consensus
AI in the public sector isn’t a quick fix. While there’s optimism about potential cost savings, real-world evidence of success is limited. The impact on public service quality and delivery must be considered carefully. Decisions about AI use should be informed by political leadership, independent expertise, and public input.
Engaging the public, especially those using and working in public services, is crucial. Participatory policymaking and consulting with worker representatives can help ministers and senior public sector leaders understand how AI will affect roles and services.
The new government should explore AI’s potential benefits while handling it with care, skill, and precision. By addressing knowledge gaps and market imbalances, they can build a foundation for a thoughtful approach to AI in public services, aiming for sustained public value.
This article is based on one by the Ada Lovelace Institute and has been adapted under a Creative Commons licence (CC-BY 4.0). Read the original article: AI in the public sector: white heat or hot air? (2024) here.
