Thank you for trying to be constructive. Should your recommendations start with an administration that supports R&D, higher education, global partnerships, and other sources of expertise?
Shouldn’t that expertise be primarily driven to work for government to support a broad public welfare? No pay premium substitutes for a desire for public service.
Don’t we need an administration that supports Federal and State institutional capacity to evaluate and execute ideas?
Yes, all this applies to AI. But also policy for economics, diplomacy, military and information. The debris of its absence surrounds us. AI policy is but one example.
Sharp piece, Dean. Restrict the best models to an approved few, and you do much more to the US democracy, the US economy, the AI investment math, and the sovereign AI thesis than delaying a launch. And all elements compound nonlinearly.
What if the hardest disagreements around frontier AI are not technical at all?
If different societies fundamentally disagree about what counts as “safe,” “acceptable,” or even “good,” can better auditing ever resolve that?
Brian Merchant has argued that AI conflicts often resemble the Luddites: the deepest disputes are ultimately about power rather than machines.
Could AI safety face the same problem, where technical standards cannot fully resolve disagreements because the disagreement is ultimately about legitimacy?
On the independent ecosystem, I didn't read all of it but METR's Frontier Risk Report seemed like a promising start and an example of what could be done.
I would also expect a ton of money to flow into independent safety orgs as soon as the OpenAI/Anthropic IPOs happen (especially Ant). So even without government promotion, these groups may soon become larger and more capable anyway.
I largely agree with you on points 12-33. Where I do not "agree" with you, it is likely a matter of semantics and strategy, rather than the recognition of the point, or a different opinion on the tactics rather than results.
You own Point 34 entirely.
Regarding Point 7: What did you fail to do? Where or what is your fault? How does your fault inform next steps?
Thank you for trying to be constructive. Should your recommendations start with an administration that supports R&D, higher education, global partnerships, and other sources of expertise?
Shouldn’t that expertise be primarily driven to work for government to support a broad public welfare? No pay premium substitutes for a desire for public service.
Don’t we need an administration that supports Federal and State institutional capacity to evaluate and execute ideas?
Yes, all this applies to AI. But also policy for economics, diplomacy, military and information. The debris of its absence surrounds us. AI policy is but one example.
Sharp piece, Dean. Restrict the best models to an approved few, and you do much more to the US democracy, the US economy, the AI investment math, and the sovereign AI thesis than delaying a launch. And all elements compound nonlinearly.
What if the hardest disagreements around frontier AI are not technical at all?
If different societies fundamentally disagree about what counts as “safe,” “acceptable,” or even “good,” can better auditing ever resolve that?
Brian Merchant has argued that AI conflicts often resemble the Luddites: the deepest disputes are ultimately about power rather than machines.
Could AI safety face the same problem, where technical standards cannot fully resolve disagreements because the disagreement is ultimately about legitimacy?
On the independent ecosystem, I didn't read all of it but METR's Frontier Risk Report seemed like a promising start and an example of what could be done.
I would also expect a ton of money to flow into independent safety orgs as soon as the OpenAI/Anthropic IPOs happen (especially Ant). So even without government promotion, these groups may soon become larger and more capable anyway.
Thank you for this article and protol draft.
I largely agree with you on points 12-33. Where I do not "agree" with you, it is likely a matter of semantics and strategy, rather than the recognition of the point, or a different opinion on the tactics rather than results.
You own Point 34 entirely.
Regarding Point 7: What did you fail to do? Where or what is your fault? How does your fault inform next steps?