AI network

The March AI Network was an opportunity for local government colleagues to share experiences of exploring and implementing Copilot in councils. The meeting provided an opportunity to discuss the benefits and challenges experienced as well as the impact of Microsoft dominance in the market. A short update on the LGA’s AI survey’s early findings was provided, as well as usual partner updates from Socitm, LOTI and iNetwork on their AI activity.

Representatives from the Local Government Collaboration and Innovation Copilot Subgroup joined to share use cases, and the evolving role of the 280+ strong group who share insight specifically on Copilot, its opportunities & use cases, and how they aim to build a collective voice in working with Microsoft. Various councils discussed implementation challenges and successes, business case development, potential benefits in social care, and pushes for reduced licensing costs. Colleagues heard from a district council, reflecting the need for clear business cases in leaner councils with fewer resources and smaller budgets when considering procuring Copilot or other AI tools.

Perspectives on the role of Copilot and generative AI in health and social care were also explored, where the potential of Copilot to save time and improve efficiency must be balanced with the need for caution when working with vulnerable individuals and sensitive information. The importance of governance and the need for human oversight to ensure the accuracy and safety of AI-generated information was emphasised. A collective statement outlining the need for responsible use of generative AI in adult social care was shared.

Professor Nigel Crook, Professor of AI and Robotics at Oxford Brookes University discussed “what’s under the hood” of Generative AI and Large Language Models, explaining advancements from early projects through to modern technologies like GPTs. This was to support members to make informed choices in relation to risks when using generative AI tools and considering approaches to governing their use. 

In discussion, Professor Crook confronted concerns about hallucinations, detecting AI generated outputs, the importance of robust contract clauses between councils and suppliers, and how councils can mitigate the risk of harm in GPT-based bots. 

A closed sector discussion on local government considerations for AI identified needs for better supplier engagement tools, regional knowledge sharing, addressing supplier lock-in, and staff skill gaps.

This meeting explored how the ICO and EHRC’s existing guidance on compliance with data protection law and public sector equality duty can be applied to the principles ‘Transparency, Explainability, Fairness, Equality’.
 

The Information Commissioners Officer (ICO) joined the call, to explain their updated guidance on fairness in data protection & AI to colleagues. This covered: defining data fairness, algorithmic fairness, bias in the AI lifecycle, exploitation vs exploration of data, and the importance of transparency in explaining AI decisions.
 

The Equality and Human Rights Commission (EHRC) joined to discuss their role in tackling discrimination in AI & digital services covered: public sector equality duty (PSED) and AI, incorporating equality into AI procurement, challenges and considerations with explainability, equality impact of AI on workforce and the procurement life cycle.

Five councils joined the network to discuss varying approaches to the use of Large Language Models 4 in councils, and broader strategic approaches to the deployment of AI. Participants covered establishing governance boards, involving senior leadership, and focusing on the needs of the community, with a focus on transparency and ethical practices. Others are also exploring AI implementation with a cautious but curious mindset, focusing on building public trust and ensuring organisational awareness, with an emphasis on the need for collaboration with suppliers early on.

Discussions showed that the sector faces challenges in adopting AI with fears of falling behind, high costs, and lack of clarity on vision and values. In the current local government climate, challenges in finding suitable suppliers and ensuring transparency in the procurement process have the potential to hinder AI adoption. There are also workforce concerns including the need for consistent skills, training and transparency in the transition to AI. Developing capacity and capability in understanding data and procurement is also essential for effective AI adoption in local government.

The LGA’s Cyber, Digital and Technology team shared plans for a general support offer over the coming year, including an AI readiness sector survey, “AI Unpacked” explainer videos, and support for councils in the governance of digital transformation within councils, with a focus on AI. Partners including Socitm, LOTI and iNetwork also joined to update colleagues.