AI and the Reliance on Decision-Making Tests in the Income Allocation Rules

Abstract

Artificial intelligence (AI) will have a transformative impact across many aspects of the world, including important implications for tax systems. This article addresses one aspect of that impact, namely the challenges created where AI is deployed by an MNE to deliver optimized decision-making in the conduct of its business. This is significant because the existing tax rules that govern the allocation of MNE business income place considerable emphasis on the location where certain decision-making functions are discharged. Those tax rules do not expressly require that economically significant decisions must be taken by natural persons, but are structured around the implicit assumption that such decisions are attributable to identifiable decision-makers whose activities can be located within a particular jurisdiction. This raises challenging questions about the treatment of AI decision-making under the income allocation rules. This article explores these questions, addressing the nature of AI decision-making and analysing the implications of such decision-making for the purposes of the income allocation rules. It argues that AI exposes a deeper issue: the extent to which existing decision-making tests depend on identifiable and locatable decision-makers.