By James Carson
Artificial intelligence (AI) has come a long way since computer pioneer Alan Turing first considered the notion of ‘thinking machines’ in the 1950s. More than half a century later, advances such as natural language processing and translation, and facial recognition have taken AI out of the computer lab and onto our smartphones. Meanwhile, faster computers and large datasets have enabled machine learning, where a computer imitates the way that humans learn.
AI has already had important impacts on how we live and work: in healthcare, it’s helping to enhance diagnosis of disease; in financial services AI is being deployed to spot trends that can’t be easily picked up by conventional reporting methods; and in education, AI can provide learning, testing and feedback, with benefits both to students and teachers. And now, intelligent automation is being adopted by local government.
AI goes local
A decade of austerity has left local councils struggling to ‘do more with less’. The Covid-19 pandemic has presented additional challenges, but has also accelerated efforts by local government to find digital solutions.
AI offers local authorities the benefits of streamlining routine tasks and processes, freeing up staff to focus on higher value activities which deliver better services and outcomes to citizens. Intelligent automation could also have important economic impacts. IPPR has estimated that AI could save councils up to £6bn in social care costs.
When it comes to system and data updating, intelligent automation really comes into its own. From managing council tax payments to issuing parking permits, there are now digital solutions to the many task-driven processes that are such a major part of local government’s work.
Many local councils are also exploring the application of chatbots or virtual assistants. These technologies enable customer services to provide automated, human-like answers to frequently asked questions on subjects as varied as waste management, street lighting and anti-social behaviour. The time and cost savings from this kind of digital solution can be substantial. Newham Council in London deployed a multilingual chatbot to answer residents’ questions. Within six months, the technology had answered 10,000 questions, saved 84 hours of call time and generated cost savings of £40,000.
The challenges of AI in local government: getting it right
Earlier this year, a report from the Oxford Commission on AI and Good Governance identified the major challenges facing local authorities when considering AI.
Inaccurate or incomplete data can delay or derail an AI project, so it’s vital that data quality issues are addressed early on. The report highlighted a project where one local authority explored how predictive analytics might be used to help prioritize inspections of houses in multiple occupation (HMOs). Predictive analytics involves the use of historic data to predict new instances. But in this case the challenges of cleaning, processing and merging the data proved too intractable to produce successful predictions.
Another important step for local authorities is to clearly define the objectives of an AI project, providing a clear vision of the outcomes, while managing expectations among all affected stakeholders – especially senior managers. The report points to a successful project implemented by Manchester City Council which developed an integrated database that allowed them to automate record searches and build predictive tools. The project had a clearly stated aim of identifying troubled families to participate in the government’s payment-by-results programme. This approach gave the project a specific focus and an easily measurable assessment of success.
It’s also important for local councils and technology suppliers to work together, ensuring that suppliers are aware of local contexts, existing data and processes. At the same time, making full use of in-house expertise can help AI technologies work better in a local government setting. The Oxford Commission report explains that after the disappointing results from the previously mentioned HMOs project, in-house data scientists working in one of the participating local authorities developed their own solution.
Sometimes, councils will discover that AI is a good fit in some parts of their work, but doesn’t work in others. In 2019, Oxford City Council explored whether chatbots could help solve design problems in some of their services. The council found that, while waste and recycling enquiries could be easily handled by a chatbot, the complex nature of the planning service would have made it difficult to remove humans from the conversations taking place in this setting. That said, another council has found it possible to develop a chatbot for its planning applications.
At the same time, digitalisation is compelling councils to adjust to new ways of working, something discussed in a Local Government Association presentation by Aylesbury Vale District Council.
The future of AI in local government
Since we last looked at this subject, local government involvement in AI has increased. But there are still important governance and ethical arrangements to consider so that AI technologies in public services can achieve benefits that citizens can trust.
The Oxford Commission report set out a number of recommendations, including:
- minimum mandatory data standards and dedicated resources for the maintenance of data quality;
- minimum mandatory guidance for problem definition and project progress monitoring;
- dedicated resources to ensure that local authorities can be intelligent consumers and capable developers of AI;
- a platform to compile all relevant information about information technology projects in local authorities.
Three years ago, MJ magazine described AI as a ‘game-changer’ for local government. The potential benefits are clear. AI can generate labour and cost savings, but also offers the promise of reducing carbon footprints and optimizing energy usage. But while residents may welcome greater efficiency in their local councils, many will have concerns about data privacy, digital inclusion and trust in the use of public data.
At its best, artificial intelligence will complement the services provided by local authorities, while ensuring that the all-important element of human intelligence remains at the heart of local government.
Further reading: more on digital from The Knowledge Exchange blog