In recent years, public policy makers leverage large amounts of policy related digital data that are generated through different channels (e.g., e-services, social media) to realize a shift towards data-driven evidence-based policy development. The advent of Machine Learning (ML) and Artificial Intelligence (AI) holds the promise to facilitate and accelerate this shift, through easing and automating the processing of large datasets, while helping policy makers to identify unique, yet previously hidden, policy development insights. Nevertheless, the use of AI for public policy development is also associated with significant technical, political and regulatory challenges.

This paper discusses these challenges and suggests a range of technical and technological solutions for overcoming them. The latter solutions include a reference architecture and a blueprint data mining process for AI-based policy making, along with AI algorithms that alleviate AI bias and boost transparency and explainability. Moreover, the paper presents the practical validation and use of these technological building blocks in real-life public policy making cases.

View Full-Text