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

They’ll move to havens with lax standards to dodge regulation

Dr. Amanda Askell 19, Research Scientist on the Policy Team at OpenAI, PhD in Philosophy from New York University, Dr. Miles Brundage, AI Policy Research Fellow with the Centre for the Governance of AI at Future of Humanity Institute, PhD in Human and Social Dimensions of Science and Technology from Arizona State University, and Gillian Hadfield, Schwartz Reisman Chair in Technology and Society, Professor of Law, and Professor of Strategic Management at the University of Toronto, PhD in Economics from Stanford University, “The Role of Cooperation in Responsible AI Development”, Computers and Society, https://arxiv.org/abs/1907.04534

1.4.4 The difficulty of constructing effective AI regulation

There is currently little in the way of AI-targeted regulation, including government regulation, industry self-regulation, international standards, and clarity on how existing laws will be applied to AI (see note 13). Well-designed regulatory mechanisms can incentivize companies to invest appropriate resources in safety, security, and impact evaluation when market failures or coordination failures have weakened the other incentives to do so. Poorly-designed regulation can be harmful rather than helpful, however. Such regulation can discourage innovation (Heyes, 2009) and even increase risks to the public (Latin, 1988).

AI regulation seems particularly tricky to get right, as it would require a detailed understanding of the technology on the part of regulators.21 The fact that private AI companies can generally relocate easily also means that any attempt to regulate AI nationally could result in international regulatory competition rather than an increase in responsible development.22 Regulation that is reactive and slow may also be insufficient to deal with the challenges raised by AI systems. AI systems can operate much faster than humans, which can lead to what Johnson et al. (2013) call ‘ultrafast extreme events’ (UEEs) such as flash crashes caused by algorithmic trading.23

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