As a wave of digitisation sweeps Europe, algorithms, and artificial intelligence (AI) rapidly become an integral part of our daily lives. Public authorities thus use algorithms to optimise their processes, improve the accessibility to their services, or manage public spaces. This development stresses the need for ensuring that algorithms in public services must respect the same principles of responsibility, accountability, transparency, and security as in all other public goods and services in which personal data are processed. Public registers are one of the solutions.
In the cities of Amsterdam and Helsinki more than thirty AI projects are taking place at the moment, among other things for improving and controlling parking, reducing house rental fraud and using chatbots to improve public services. This has led to a new initiative, public AI registers.
The AI registers contain information about applied datasets, the model architecture and the operational logic in the system, how non-discrimination and realisation of equality is achieved, how human oversight during the use of the service is secured, and which risks are related to using the system.
The purpose of AI registers is to empower citizens with knowledge of algorithms and up-to-date information about how algorithms support decisions and affect their lives. On this basis they are better prepared to voice their opinions on public sector activities and participate in the debate about the development in AI and automatization in their local areas.
Transparency and Feedback
As such AI registers represent a window into the AI systems set up and used by a public authorities. The registers provide information which helps everyone to get acquainted with the overviews of AI systems or to examine them in more detail. This opens for inclusion of stakeholder interests and participation and as a channel to provide feedback to planned or running systems. For journalists, scientists, politicians, civil society organisations and activists, the AI register can become an invaluable help while ensuring public insight and control over societal impacts of AI.
Especially, the feedback mechanism enables government organisations to comply with citizen and stakeholder requests on the workings of algorithmic systems. It also helps document the decisions and assumptions that were made in the process of developing, implementing, managing, and perhaps dismantling an algorithm.
Hereby, an AI register contributes to fulfilling data ethical demands for transparency and explainability.
Good AI Governance
For civil servants, an AI register represent a relevant tool for solid AI governance in development and operation processes. It will help teams create and maintain documentation systematically across all their projects, regardless of the technologies and partners involved.
From a data ethical perspective, AI registers represent a relevant vehicle for transforming data ethics into practice through a participatory approach to building and operating algorithms in public administration. In other words, a sound pathway to trustworthy AI in our cities.