There is a consensus that AI will fundamentally transform modern economies and societies, and be applied across numerous day-to-day activities of work and life. While the expectations are large, there are also concerns. For regulators, in particular, it is crucial to understand the different dimensions of AI, what needs to be regulated and why. The need for regulation could arise from the risk of discrimination as well as safety and privacy concerns, also raising the question of accountability in case of undesired outcomes.