This article is intended to inform policy-makers in their decisions in deciding their approach to governance of AI. The authors argue that whether AI governance should be centralized depends on the depends on exact design of central body and that current efforts must be continuously reviewed and evaluated to see if they are self organizing. The authors highlight that centralization can be beneficial for institutional efficiency and increasing participation. However, they also bring to light the problems of centralization such as the slow process of establishing an institution and the brittleness of institutions even after coming into existence. Additionally, in considering the ‘breadth vs depth dilemma’, seeking centralization can heighten the barrier, limiting participation. Lastly, the authors introduce the consideration for decentralized governance’s strength in being able to support forum shopping which could either help or hurt governance.
