Free beta access. 50 free AI tokens each week. Optional top-ups available. See what is included
Opening Word document…

The creates an IIGO which solves best

Erdélyi and Goldsmith ‘18 (Olivia, faculty of Law at University of Canterbury and Judy, faculty at University of Kentucky Department of Computer Science, “Regulating Artificial Intelligence Proposal for a Global Situation, Published February 2018, Accessed 7/13/2022, https://dl.acm.org/doi/pdf/10.1145/3278721.3278731)

International institutions are the prevalent vehicles of international cooperation in our interconnected world. When a critical mass of states and/or non-state actors feel that transnational cooperation is necessary to solve a problem that is impossible to tackle by isolated national measures, they establish a new IGO or NGO for that particular purpose. Based on legal and international relations definitions in circulation — see [9, 14, 18] — we define an IGO as a formal entity (1) established by an international agreement governed by international law; (2) with at least three (sometimes two) members — typically states but increasingly also IGOs; and (3) having at least one organ with a will distinct from that of its members. FIGOs’ organizational purpose is laid down in a binding international agreement such as a treaty or a formal legal act of another IGO, their membership is clearly defined in the founding legal act, and they have a permanent and significant institutionalization in place. By contrast, IIGOs operate based on an explicitly shared, but informal expectation about purpose, their membership is not always clear, as members are explicitly associated but only by non-legal mutual acknowledgment, and they do not possess any significant institutionalization. NGOs differ from IGOs in that they are not created by treaty — meaning they are governed by national rather than international law — and their membership is made up of non-state actors. Given the severity and global nature of AI’s anticipated impact on humanity, we expect it to join the long line of issue areas requiring interstate cooperation, raising the question of establishing an IGO at some point in the future. Against this background, we propose the creation of the IAIO as a new IGO, which could initially serve as a focal point of policy debates on AI-related matters and — given sufficient international support — acquire increasing role in their regulation over time. We start by determining the degree of desired institutional formalization by examining, in turn, the six above elaborated tradeoffs in relation to AI.

Binding commitment vs. flexible cooperation arrangements: As pointed out earlier, AI will fundamentally transform human society worldwide. Since this process of transformation is likely to be inescapable for any single state, states will probably wish to cooperate sincerely. Also, violations will be difficult to detect as keeping pace with technological innovation will require considerable technical expertise and capacities, presumably exceeding especially weaker countries’ capabilities and evoking severe power asymmetries. While apart from this latter circumstance, these factors speak for hard legal commitments, it must be kept in mind that AI research and AI-human interactions are relatively young phenomena and their novelty severely restricts our ability to anticipate the spectrum and extent of the impending changes, let alone the dimension of the problems they will raise. Many AI instantiations encroach on our most basic rights, pose an existential threat, or bring up profound ethical and social questions, not to mention that they will utterly and completely upset our legal system. So, we are looking at heated debates among radically diverse parties over a variety of uncertain issues, which may change in rapid and currently unimaginable ways — conditions that, based on past experience, do not exactly favor international consensus. Therefore, we need all the flexibility we can get to acquire familiarity with the issues at hand, sort out differences, and establish common ground, before we can contemplate drawing up a more binding framework for cooperation. Delegation/high sovereignty costs vs. state autonomy/low sovereignty costs: Weaponized AI technologies and certain data mining practices are clearly relevant for national security. As this is a sensitive issue area involving particularly high sovereignty costs, at least initially, states will show reluctance to give up and delegate decisionmaking authority to the IAIO. On the long run, however, powerful collective oversight and enforcement mechanisms will probably be indispensable in order to curb incentives for violations and opportunistic behavior, which should otherwise be high in light of the major shifts in international power constellations triggered by changes in countries’ competitive positions. Also, domestic AI policies will produce significant externalities, affecting other countries. Based on this analysis, it is hard to escape the conclusion that a highly institutionalized organization with binding legislative, dispute resolution, and enforcement authority would be better suited as new international AI regulator. Nevertheless, the political reality remains that until sufficient clarity is reached on the IAIO’s precise purpose, membership, the issues to regulate, and the broad directions to follow, international consensus supporting such a high degree of institutionalization is off the table. Collective control of information vs. close control of information: History shows that states are generally cautious about sharing information on fate-changing technologies, which speaks for close control of information with respect to AI. However, if and when we manage to gather consensus for hard legal commitments (e.g., treaty on certain AI applications), we will probably need to be more forthcoming with certain information to ensure compliance with those instruments. This is again a strong argument in favor of starting cooperation on AI regulation in a softer institutional framework and using soft law instruments, although a move towards harder legalization seems to be desirable over time.

Sign in to recut, update, or save this card.