Trading AI: Economic Interests, Societal Choices and Multilateral Rules

Abstract

After technology’s decade of disillusion, societies confront the decade of decision as Artificial Intelligence (AI) applications gain power and use cases proliferate. As a general purpose technology with applications that potentially touch on virtually every human endeavour, the integration of AI into social and economic frameworks will likely vary systematically and idiosyncratically across jurisdictions based on different economic interests and societal choices made in a context of limited experience, highly imperfect information, and at best a rudimentary understanding of the complex feedbacks that will be unleashed as the integration of AI proceeds. From a trade perspective, national regulations on AI will constitute non-tariff measures that could effectively prohibit market access abroad. The trading system has some experience in dealing with risks related to new technologies, including the use of scientific evidence, factors to be considered in assessing risk, role of international standards, and the precautionary principle. However, this experience also highlights the challenges that will likely be encountered in sorting out legitimate derogations from the freedom of commerce that international economic law defends from ordinary protection. If dealing with sanitary and phytosanitary issues was tough, given the much broader range of interests that it touches, AI promises to be commensurately tougher.

Keywords: artificial intelligence, AI, machine learning, ML, international trade, WTO, SPS, dispute settlement, precautionary principle, national security, machine knowledge capital

JEL Classification: O32, O33, O38, K20, K21, K24