AIThe case for refocusing a superintelligence bill on AI investment
One commenter welcomes the bill's audit powers and approval requirements for very large models, but says its licensing threshold is too low and its rules on legitimate grounds for rejecting licenses lack precision.
TLDR
A commenter praises a proposed superintelligence bill for taking catastrophic risks and disempowerment seriously. They describe sweeping government authority to audit AI models and companies, approval requirements before development and deployment of very large models, and a de facto licensing regime. But they consider the licensing threshold too low and the grounds for rejecting licenses insufficiently precise. Their preferred rewrite would remove the focus on banning what they call vaguely defined artificial superintelligence, instead targeting the investment, profits and computing power fueling the AI industry.
The case for refocusing a superintelligence bill on AI investment
One commenter welcomes the bill's audit powers and approval requirements for very large models, but says its licensing threshold is too low and its rules on legitimate grounds for rejecting licenses lack precision.
TLDR
A commenter praises a proposed superintelligence bill for taking catastrophic risks and disempowerment seriously. They describe sweeping government authority to audit AI models and companies, approval requirements before development and deployment of very large models, and a de facto licensing regime. But they consider the licensing threshold too low and the grounds for rejecting licenses insufficiently precise. Their preferred rewrite would remove the focus on banning what they call vaguely defined artificial superintelligence, instead targeting the investment, profits and computing power fueling the AI industry.
