学术文献库

Artificial intelligence is already being applied in and impacting many important sectors in society, including healthcare, finance, and policing. These applications will increase as AI capabilities continue to progress, which has the potential to be highly beneficial for society, or to cause serious harm. The role of AI governance is ultimately to take practical steps to mitigate this risk of harm while enabling the benefits of innovation in AI. This requires answering challenging empirical questions about current and potential risks and benefits of AI: assessing impacts that are often widely distributed and indirect, and making predictions about a highly uncertain future. It also requires thinking through the normative question of what beneficial use of AI in society looks like, which is equally challenging. Though different groups may agree on high-level principles that uses of AI should respect (e.g., privacy, fairness, and autonomy), challenges arise when putting these principles into practice. For example, it is straightforward to say that AI systems must protect individual privacy, but there is presumably some amount or type of privacy that most people would be willing to give up to develop life-saving medical treatments. Despite these challenges, research can and has made progress on these questions. The aim of this chapter will be to give readers an understanding of this progress, and of the challenges that remain.