ABSTRACT
To design effective policies to govern AI, it is useful to consider some key variables such as the distribution of benefits and costs from AI across society, the breadth of access to the most capable AI systems, and whether the capabilities of AI will be concentrated in a few highly powerful systems or more diffused. This chapter starts by reviewing various scenarios with regard to these topics. Policies on AI should also take account of the speed of AI's effects on the economy, which are discussed in the next section. The latter part of the chapter discusses how various stakeholders can contribute to good governance and effective risk management, including companies developing AI, national governments and international bodies. Some forms of international cooperation may be needed to avoid a race to the bottom on safety standards, but there are difficult questions to be addressed around such cooperation should be designed and incentivised.
