From statisticians to data scientists – The emergence and limitations of AI applications in public services

Development of Artificial Intelligence application requires both technological and human resources. The requirement of a skilled computational workforce has made AI applications closely linked with the emerging data science profession. In this paper I draw from sociology of science and technology to discuss the application of AI and data science tools into the design and delivery of public services. The perspective of data scientists is contrasted with the tools and practices of more established quantitative professions, such as statisticians, economists, and management analysts. The findings draw from 39 interviews with experts of quantitative and computational tools in the UK non-profit sector. The UK non-profit sector is closely linked with the UK government through contracting and collaboration, making it central to design and delivery of public services. The findings suggest that AI tools and data science applications have so far only had a minor impact on non-profit work, and is unlikely to have a major impact in the near future. In projects designed to explore new possibilities of computational tools in the non-profit sector, the signature tools of data scientists often end up not being applicable. The interviews suggest that the quantitative tools and practices currently used in the non-profit sector are supported by computational infrastructures that have become a standard part of public service design and delivery over the decades. The emerging data science profession, in contrast, draws mostly from computational infrastructures that originate from computer science and digital commerce. While data scientists as professionals can be transported to the non-profit sector, they still lack the computational environments that has enabled the recent advances in computational technology. The findings thus suggest that the potential of AI in public service design and delivery is currently restricted to a limited set of applications. New developments are better described as incremental improvements to existing computational practices. Nevertheless, the emergence of a special data science professional identity and skill set stands to have more direct implications to the non-profit sector.