What Altman proposed in May 2023 to shift attention from present harms to future systems
Karen Hao's framework, from Empire of AI, pp. 548 to 619.
Licensing agency
Create an agency to license models above a threshold of capability
Safety standards
Create a set of standards for measuring dangerous capabilities
Independent audits
Require independent audits to check compliance with those standards
From chapter: Index, pp. 548–619
The index lists names and topics alphabetically with printed page numbers, from Abbeel and acceleration risk to Zuckerberg. It covers people such as Altman, Sutskever, Murati and Brockman and topics such as AGI, scaling laws, RLHF, data centres and Scale AI. It is a navigation aid with no argument.
