Ai Governance And Regulation: Emerging Approaches

Abstract

Artificial intelligence (AI) systems are conquering the world and facilitate the shift to an automated future where many work processes become replaced by machines. Significant productivity gains will be reaped from this digital transformation. However, the change to data-driven AI decision-making poses various risks, for instance, can heighten bias/discrimination; removes subjective nuances and discretion which form part of human decision making; creates opacity; leads to unforeseeability/unpredictability; and can centralise control and risk that democracy becomes replaced by technocracy. The regulation and governance of AI is therefore a pressing concern, and four regulatory approaches are emerging: The European risk-based approach, the UK sectoral approach, a human rights approach, and a hybrid approach. These approaches seek to promote innovation and safeguard against risks to varying degrees. The risk-based and sectoral approaches reflect the drive of the EU and the UK to become global leaders in the AI sector since they have opted for an innovation friendly regulatory framework. A theory is proposed for adopting the appropriate regulatory approach. It is emphasised that human rights can play an important role for AI alignment and that even innovation-focused approaches should give more space to human rights.