Richmond upon Thames Council is set to trial the use of artificial intelligence (AI) to detect potholes, as part of a broader effort to improve efficiency, save money, and provide better outcomes for residents. The decision was made at a Policy and Performance Review Board (PPRB) meeting on Tuesday 21 October 2025, where the board discussed the potential of AI solutions within the council, as detailed in the Use of AI report FINAL.
Sam Olsen, Executive Director Change and Innovation, presented a report outlining the current and future uses of AI within the council. He explained that AI encompasses systems capable of performing tasks requiring human intelligence, such as reasoning and problem-solving. While AI has been around for decades, recent advancements have increased its adoption.
The council is already using AI in various ways, including spam filters and spell checking. A trial of Microsoft 365 Copilot, an AI assistant integrated into Microsoft Office applications, was conducted in 2024. In Adult Social Care, a pilot of an AI solution to assist in generating assessments for unpaid carers is underway. Richmond has also been selected to trial an AI meeting assistant called Minute.
Looking ahead, there are several potential AI developments, including the Digital Blueprints initiative, AI voice agents, and automating routine processes. One specific example highlighted was the use of cameras on refuse collection vehicles to track the size and state of potholes, predicting potential road defects before they arise. This system uses computer vision to identify, measure, and classify road defects.
Olsen noted that while AI offers significant opportunities, deploying AI solutions comes with challenges and risks, including data security, legal compliance, and the need for staff training. To address this, the council is providing training to staff to understand both the capabilities and limitations of AI tools, including how to formulate prompts for desired outcomes and interpret the results from predictive models. The training also aims to develop general awareness of AI, including recognizing issues like model bias and privacy concerns. An online community called 'Copilot Navigators' was created to support the development of AI skills such as prompt writing and to enhance understanding of AI applications among staff members.
To ensure the ethical use of AI, the council's Generative AI Policy mandates that staff only use approved, secure AI platforms and never upload personal or sensitive data into untrusted AI tools. The council also ensures that any AI tool or service used by the organisation complies with laws and upholds ethical standards, including data protection laws, equality legislation, freedom of information, and human rights.
Councillor Nicholson raised concerns about the accuracy of AI-generated answers, to which Olsen responded:
You should never expect a piece of technology to be your final fail safe on the right answer for anything, and that AI is there to help collate large amounts of information, not to answer questions for the council.
Councillor Cardy raised concerns about data security within the council, and the risk that AI could be used to create information that gets past security measures. The council requires vendors to detail their AI systems and how possible AI bias is tested and mitigated to provide transparency in AI decision-making. The council's processing notices will be updated to reflect the potential use of AI solutions in advisory or assisting roles within decision-making.
The success of AI initiatives will be measured by time savings, cost savings, and improved service delivery. The council is also looking at using AI to automate routine processes, support data analysis, and facilitate enhanced customer engagement.
The board agreed to note the report and requested a follow-up report in a year's time.
