Assessing the social return of AI in UK local government: a community-centred approach

As the financial crisis of UK local government unfolds, local authorities are increasingly turning to systems powered by Artificial Intelligence (AI) as a means of delivering frontline services. For example, in 2023 Derby City Council began piloting its strategy for harnessing AI to transform service delivery, deploying AI-powered chatbots to handle enquiries from residents. As such, the issue of how local government procure AI systems is now a live issue that requires attention, particularly given the predominance of narratives which emphasise the inevitability of AI-powered solutions to issues of governance and service delivery. In the UK, it is a statutory requirement that local authorities consider the so-called social value of any service procured. In principle, this social value requirement should provide ample opportunity to assess the social return on investment offered by AI services. However, due to weaknesses in existing social value frameworks, evaluation is limited. This paper, using Derby City Council as an illustrative case study, will outline why the predominant UK frameworks for assessing social value (the ‘social value model’ and the ‘National TOMs’ approach) are inadequate when considering AI technologies. In doing so, it will posit a community-centred alternative to existing social value assessments as a means of foregrounding the agency of user-communities in the evaluation of proposed AI solutions. Such an approach not only allows for a clearer assessment of the social outcomes of AI technologies throughout the delivery chain, it also places the voices of underrepresented communities at the forefront of social value evaluation. Doing so provides an opportunity to resist the framing of AI as an inevitable solution to existent problems of local government service delivery.