Driving Public Good with AI

The United Kingdom needs to build new homes, but traditional planning applications are complicated and time-consuming, with complex paper documents requiring processing and verification against local planning restrictions. Extract, a new system built by the UK Government’s AI Incubator team, helps local government councils to convert planning documents into digital structured data, drastically reducing the time required for dealing with the roughly 350,000 planning applications filed in England every year. »The new generative AI tool will turn old planning documents – including blurry maps and handwritten notes – into clear, digital data in just 40 seconds«, explains the UK government. The new system, which is based on Google’s Gemini, is currently being tested by four councils and will be available to all councils by Spring 2026.

The stereotype goes that the public sector is slow to act when it comes to innovation, especially in terms of technology. But according to Public Sector Tech Watch, a monitor of emerging technology adoption in the European public sector, over 1,400 AI projects were underway in 38 European countries as of May 2025. Almost a third of these (29%) can be described as »general public services« – a category that includes chatbots and tools for internal office management.

The categories »economic affairs« and »public safety« rank second and third, with 18% and 16% of all use cases. Almost 600 of those projects have already been implemented and more than 600 were in the pilot stage. The countries with the most cases were Germany, the Netherlands and Italy.

»There’s a clear trend of increasing AI adoption and pilot projects in all European countries,« says Gerhard Hammerschmid, Professor of Public and Financial Management at the Hertie School of Governance in Berlin. Hammerschmid co-authored two reports on AI for the German parliament in 2019. When asked about the potential of AI to improve public services in Europe, he says he is »quietly optimistic« that the technology will bring considerable positive change. Indeed, a report published by Implement Consulting Group concludes that adopting generative AI in public administration could significantly enhance overall productivity, since the public sector accounts for 20–25% of generative AI potential in the EU. Public administrations in EU member states can enhance their productivity by 10% with generative AI and thereby unlock a EUR 100 billion opportunity. Low-risk use cases make up 15-20% of the potential.

Potential for improvement

Hammerschmid sees three areas of potential for AI in the public sector. The first relates to relationships between citizens and government, such as application forms and chatbots: »Many public services can be individualised or offered proactively«, he says. The second is office management and organisation. For Hammerschmid, this is where the greatest efficiency gains are likely to be achieved »because that’s what public sector work is mostly about«. The third area is providing support for decision-making processes such as approvals, monitoring and control. »There’s a lot of potential for improvement in this area, but it’s also the most difficult one,« he says.

Google has collaborated on projects in each of Hammerschmid’s categories. For example, the Département du Var in southern France collaborates with Google Cloud on a number of AI use cases like a website chatbot and a survey chatbot as well as automated transcription of council meetings. The Italian provincial capital Padua created a Google-powered AI voicebot to answer residents’ queries about local air pollution levels. By implementing the voicebot, the team saves 20 hours per week, with 96% of customer queries being resolved successfully. Westminster City Council in London, in partnership with Google Cloud and service provider Ancoris, has built a platform for citizens to report dumped waste and request street cleaning. The system showed 86% accuracy in automatically categorising issues.

However, harnessing the full potential of AI for public good requires a thoughtful regulatory framework. The EU’s Artificial Intelligence Act that came into force on 1 August 2024 is a key part of that regulatory environment in Europe. With the law now in the implementation phase, it will be critical to ensure that it strikes the right balance between increasing trust and ensuring innovation, including enabling rather than stifling adoption of AI by the public sector.

»I think it’s a fair worry,« says Richard Stirling, who in 2016 founded Oxford Insights, a British consultancy focused on AI adoption in the public sector. Yet Stirling points out that fears of Europe falling behind were already expressed when, in 2018, the General Data Protection Act and, in 2022, the Digital Services Act were implemented. »In both cases, what happened was the rest of the world started to follow the European Union’s lead.« That said, Stirling can also imagine that the EU AI Act could cause the EU being »treated like a second-class citizen« when it comes to rolling out AI services: »You can actually see this happening already,« he warns.

Well positioned for AI adoption

Oxford Insights annually publishes the »Government AI readiness index«, which analyses the ability of governments to harness AI. According to the latest report, European governments are well positioned for AI adoption. Western Europe ranks globally as number 2 in readiness, with an average score of 69.5, and Eastern Europe as number 4, with an average score of 57.88. This compares to an average score of 82.6 for the leading region, North America. In the national ranking, the US leads with a score of 87.03, but Western European countries dominate the top ten: France, the UK, the Netherlands, Germany and Finland.

This might seem surprising given the public perception that European countries trail the US, China and others when it comes to artificial intelligence. Stirling, who is also a »UNESCO AI Expert Without Borders«, refers to the analytical framework of his report to explain the seeming contradiction. While Western European countries tend to score well on aspects such as governmental strategic vision, regulation, IT infrastructure and flexibility to adapt, the European private sector is clearly at a disadvantage compared with that in the US.

»There’s a real opportunity in Europe to support our private sector innovation,« Stirling argues. He points to the lack of a big venture capital market as one of the root causes for the US’s lead: »Europe doesn’t have a good way for bringing capital towards large speculative bets, not like what Silicon Valley does,« Stirling says.

Still, Stirling is convinced that many European countries have already taken important steps to successfully implement AI into government processes. »The European Commission has been investing in data infrastructure across member states for more than ten years,« he says. He thinks that, going forward, rethinking procurement could play a crucial role. »There’s a rule that says you can’t talk about technology in government without talking about procurement rules. We have to design procurement systems that are flexible enough to harness the pace of change within technology.«

When it comes to designing a successful roadmap for AI adoption, Professor Hammerschmid thinks that hiring the right people will play a crucial role: »We need to bring people in who are enthusiastic about using them, and have implemented such projects in the private sector.« He also recommends focusing on small flagship projects and open-source solutions. »Once it works, you can scale these solutions and, ideally, share them all over Europe.«

AI is already creating value in the public sector in many places.

North America

USA
The Sullivan County virtual agent is an artificial intelligence-driven programme designed to immediately answer common questions on the county website. When someone clicks on the chatbot logo in the lower right of their screen, the virtual agent provides a list of topics and a space to type in questions.

USA
MIRA stands for »Medicaid Intelligent Redetermination Assistant«. It helps Medicaid
staff clear large backlogs of post-pandemic cases and was launched in 2023 with Google.

South America

Brazil
A chatbot called »Anna« helps the City of Santana de Parnaíba communicate with residents. »Anna« can handle citizen inquiries on 400 topics, from tax services to new decrees and public works.

Argentina
Prometea, an AI virtual assistant for judicial officials, predicts case solutions and provides information for assembling case files, thus freeing judicial officials from repetitive tasks and potentially streamlining the justice system.

Colombia
By adopting cloud solutions and applying data analytics powered by AI, the Social Security Management Unit significantly boosted team productivity and allowed citizens to access their social security contribution data digitally within minutes rather than days.

Africa

Ghana
The Open Buildings dataset, launched by a Google research team in Accra, uses AI and satellite imagery to pinpoint building locations in remote areas, providing critical data that helps governments understand residents’ needs to prioritise and deliver essential services, including electricity.

Nigeria/Somalia
Google developed an AI-based flood forecasting tool and leveraged it in several African states. In a pilot project in Nigeria in 2023, Google flood forecast was used to alert local communities. In Somalia, Google provided the Red Cross with inundation risk maps and daily flood forecasts.

Asia

Thailand
Thailand’s government entered a strategic partnership with Google to modernise government services and public sector delivery through AI technologies, beginning with public transportation, e-government services and big data usage.

Singapore
Smart Nation is Singapore’s overall strategy for digitising the nation, and AI is one component. The initiative leverages AI to optimise traffic flow, energy consumption and more.

Europe

Germany
In a proof of concept with Google Cloud, the German city of Wiesbaden, just west of Frankfurt, used machine-learning tools to locate potholes and other street damage by analysing videos of streets and sidewalks. With a hit rate of 94.6%, the solution has so far been extremely reliable.

Greece
The Municipality of Athens used Google Cloud to transform its citizens’ access to information using advanced keyword searches on its website and an AI chatbot with natural language recognition.

United Kingdom
The UK’s Government Digital Service upgraded the search engine on Gov.uk, the main portal for all information about the government. By using Google’s Vertex AI, it improved semantic search capabilities, with tangible results: 14% more users than before are clicking through to relevant results, according to internal analysis.

Illustrations: Ana Miminoshvili