The local authority is developing a resident focused programme of AI use cases, writes David Little, managing director of HSO
Artificial intelligence has become more accessible to local government over the past couple of years, prompting an explosion of interest in how it can be used to transform services.
But it is also stirring up fears among some employees and the public about how it will affect their lives, and creating the challenge of how councils can ensure that people feel comfortable with them using the technology.
They have to choose the right use cases to show that it provides value for the organisation and the public. This is making many cautious, looking to be early followers rather than first adopters.
But some local authorities are emerging as pioneers. Basildon Council is among those moving quickly, working with HSO to use Microsoft technology to create a resident focused approach to AI deployments that provides benefits for its citizens, staff and elected members.
Its assistant director for digital, customer and transformation, Shanna Chandiwana, shared its experience in a recent UKA Live discussion, along with myself, Microsoft UK’s director of local public services Robin Denton and UKAuthorITy publisher Helen Olsen Bedford.
Chandiwana said the initiative has come the need to deal with huge financial pressures – as faced by all local authorities – and in response to rising public expectations to use the technology, building the “next generation council”.
First steps
Basildon is beginning with three ‘trailblazer’ use cases, one being the development of an autonomous AI agent for customer services – part of a refresh of the council’s website – involving the use of sentiment analysis to provide a “humanised” approach for residents to navigate services and hand over to a human where required.
A second use case involves deploying AI agents to enable council officials to quickly find information relevant to homelessness and temporary accommodation. A third is to support HR operations through searches of internal policy and document discovery. The main thrust of each is to free up staff to use their expertise on tasks that need a human understanding.
In addition, the council has identified initial savings of £1 million, with a view to showing an early return on investment to support a scaling up of its use of AI; and acknowledged that it has to win the hearts and minds of the people involved.
“The first part of this journey isn’t sexy,” Chandiwana said. “We start talking about policy, security, data audits and thinking about readiness.
“The work with HSO has been brilliant in terms of the AI readiness assessment, because most staff have been involved in some way or know someone who has been involved in the workshops. We’ve also had a roundtable with our elected members. It’s around making sure the organisation can keep engaged.”
Managing risks
There are risks involved in an AI programme, notably those of costs and time being greater than expected, and the dangers of an organisation’s reputation being severely damaged if things go wrong.
This requires the management of expectations to make clear that AI will not perform miracles and not every project will be successful. More importantly, for services that have a deep effect on people’s lives – such as those around bereavement or domestic violence – there have to be guardrails to ensure that no key steps are taken without human oversight.
It leads to one of the key requirements that we at HSO have identified for planning a programme, to carry out a process-by-process risk assessment. If it’s a simple request for information or a routine, non-sensitive transaction it can be fully automated; but it will be crucial to keep a ‘human in the loop’ in AI deployments for more sensitive and riskier activities.
Denton emphasised the importance of this, saying that if an AI deployment is supporting safeguarding or a sensitive service, there needs to be a human in that loop reviewing whatever is coming through, and this should be regarded as a “red line”.
Another is to make sure the data is right for use with AI. This involves ensuring that it is accessible from legacy systems, looking at its quality and format to ensure it is structured in a way that can be processed by the algorithms.
As Denton pointed out, having the right data architecture in place is a step towards getting an AI deployment right, and that using a technology such as Copilot quickly shows which elements need to be fixed.
Chandiwana said that Basildon, supported by HSO, has run a data audit, assessing its systems to see which ones will be usable and what improvements are needed, and has a three-year roadmap for the decommissioning or relaunch of legacy systems.
Honesty about data
“We’ve looked at best practice, what’s right for Basildon, but I’ve never met anyone in the public sector who says their data is the best quality end-to-end,” she said, adding: “We are not going to have a pure dataverse with harmonious data, where everything can be pulled upon.
“We’ve been honest about where there are gaps, what we will and will not migrate into a clean environment, what we will get more of, and what we create as a precedent of that data, such as a single view of a customer or single record of an asset.”
She said the council will make the decisions on a “use case by use case” basis and expects to have 360 degree views of its tenants and staff within the first six months, before moving on to a complete view of all local residents.
It has also been asking its residents questions about what they would prefer from AI solutions, such as which ones should be seamlessly automated and which should come with regular updates, and is aiming to build “bite sized” case studies for each one.
“For different communities there will be different cases,” Chandiwana said. “It’s about making it real and we will continuously learn.”
Key technologies
Basildon is using the Microsoft Copilot AI assistant and Fabric data analytics platform as key technologies in the programme, working with HSO on building AI applications from them. The former makes it possible to quickly and securely upload and analyse data to support workflows; while the latter can produce an AI-ready data lake for use with a range of applications.
Chandiwana cited possibilities including obtaining new insights on issues such as areas of deprivation, needs in adult social care and for temporary accommodation.
She also emphasised the importance of an organisation asking itself questions about the ethics of any AI deployment it is planning, with a willingness to be challenged. This may sometimes create blockers, but it can also open up big opportunities over the longer term.
This all comes with the message that Basildon is producing assets that will be easily replicable in other local authorities. HSO is working with the council to ensure that others can build on its experience and make progress even more quickly in ensuring that AI is used for the benefit of their residents.
If you are a local authority just starting with AI, or looking to turn interest into value, HSO’s AI Readiness Assessment can provide you with a clear, practical plan to move forward with confidence. Click this link to arrange a conversation.
Meanwhile, catch up with the full discussion via the on-demand video below:
