But the benefits will only be achieved if we build the technology in the right way
Looking at all of this through the utilitarian lense moves the focus from individual rights to the outcomes for everyone involved, and even beyond that, because everything impacts something else, like a butterfly effect. "Pacing the frontier" could save society from AI casualties, whether that be any of the ones he mentioned.
They include the risk of losing control of AI systems, misuse of AI for cyberattacks and bioterrorism, and serious economic disruption.
Using the "rights" lens, people whose rights could be affected are employees, the government, people who regularly use AI, the overall public, and maybe even future generations who will use AI. If they are using the tool, they deserve to know the current state of it and the risks that come with it's use. People also deserve the right to their general safety, amid the risks of AI, like cyberattacks for example like the one Amodei described in his article.
Finally, it is important to note that even if we cannot achieve formal agreements, simply changing informal norms may have some value.
Society overall could benefit from slowing development through technological and medical advances. Global pacing, especially, would improve consistency across AI tools.
whose role is to verify adherence to safety practices and commitments, report incidents, and help assess the alignment of not just completed AI models but training pipelines and processes.
I think the creators of these AI tools have a responsibility to protect the rights of everyone that is using AI or may be impacted by it in some way. Evaluators tasked with ensuring the tools are working smoothly have a responsibility to be transparent about the tools as well, and to favor precision and quality over speed or efficiency.