AI in government: analysing opinions on evolving legislation on algorithmic and automated decision-making in Finland

Using AI and less sophisticated ICTs has not advanced without problems in Finland’s government and public administration. As indicated by important opinions voiced by the Deputy Ombudsman of the Finnish Parliament and the Constitutional Committee of Parliament, the full utilization of these technologies in rationalizing public services has been difficult to adjust with existing Finnish legislation on the publicness of official documents, due process, and civil servant responsibility. Issues in the use of AI and other ICTs have arisen, for instance, in administering permits to refugees residing in Finland; in surveying of taxpayers’ compliance with tax and taxation legislation; and in combining data from official registries when experimenting with reform alternatives. Conceptually, our work explores ideational tensions and conceptual shifts between AI and other ICT on the one hand, and on the other Finland’s legislative culture that frames the use of the aforementioned technologies. We support our conceptual analysis using the LDA topic modelling. We use this clustering method to identify meaningful clusters in stakeholder opinions (N~60) received by the Finnish Ministry of Justice in a public consultation process related to the preparation of new administrative legislation to settle the issues of algorithmic and other automated decision-making in government. The opinions were received in the Autumn 2020 and a working group to probe the possibilities of passing the legislation will start its work in 2021.