Artificial intelligence (AI) is beginning to transform many industries, yet its use to improve public services remains limited globally. AI-based tools could streamline access to government benefits through online chatbots or automate systems by which citizens report problems such as potholes.
Currently, scholarly advances in AI are mostly confined to academic papers and conferences, rarely translating into actionable government policies or products. This means that the expertise at universities is not used to solve real-world problems. As a No10 Innovation Fellow with the UK government and a lecturer in spatial data science, I have explored the potential of AI-driven rapid prototyping in public policy.
Take Street.AI, a prototype smartphone app that I developed, which lets citizens report issues including potholes, street violence or illegal litter dumping by simply taking a picture through the app. The AI model classifies the problem automatically and alerts the relevant local authority, passing on the location and details of the issue. A key feature of the app is its on-device processing, which ensures privacy and reduces operational costs. Similar tools were tested as an early-warning system during the riots that swept the United Kingdom in July and August 2024.
AI models can also aid complex decision-making — for instance, that involved in determining where to build houses. The UK government plans to construct 1.5 million homes in the next 5 years, but planning laws require that several parameters be considered — such as proximity to schools, noise levels, the neighbourhoods’ built-up ratio and flood risk. The current strategy is to compile voluminous academic reports on viable locations, but an online dashboard powered by AI that can optimize across parameters would be much more useful to policymakers.
How to track the economic impact of public investments in AI
This is the key insight from my stint in government: public officials are often interested in deliverable products or demonstrable solutions, whereas academics are trained to funnel new knowledge into papers. Existing mechanisms to bring academic expertise into government, such as secondment opportunities and exchange programmes, do not adequately address the factors that limit collaboration. A more effective approach would be to frame research in practical, solution-oriented terms.
There are several measures that could strengthen collaboration between academia and public institutions, particularly in emerging domains such as AI, to enhance the delivery of public services.
Academics can ease into a product-focused mindset through prototyping, where the risk is low. The aim is to go beyond conventional research-dissemination practices to ensure that findings are accessible and can be applied in government settings. This involves translating complex data, models and insights into user-friendly digital tools. For instance, imagine that a university research team has developed an AI model capable of predicting which areas are at high risk of flooding. As well as publishing the findings in academic journals, the team could create a digital tool — an interactive platform — to display the AI insights for government officials in a way that makes them actionable.
