While the USSR was indeed able to exfiltrate secrets from Los Alamos to speed up its nuclear program, it took a few more years for it to actually develop a nuclear weapon.
Russell (and we don’t necessarily agree here) argued that the US could have established a monopoly on nuclear development through nuclear coercion. That strategy doesn’t have anything to do with preventing espionage.
Once the genie is out of the bottle, it doesn’t matter, does it? Much of China’s current tech achievements began with industrial espionage. You can’t constrain a game-changing technology while excluding espionage as a factor.
It’s exactly the same issue with AI.
While you have an interesting theoretical concept, there’s no way to derive a strategy from it that would lead to AI safety that I can see.
The idea for this particular theory of victory is that, if some country (for example, the US) develops TAI first, it could use TAI to prevent other countries (for example, China) from developing TAI as well — including via espionage.
If TAI grants a decisive strategic advantage, then it follows that such a monopoly could be effectively enforced (for example, it’s plausible that TAI-enabled cybersecurity would effectively protect against non-TAI cyberoffense).
Again, I’m not necessarily endorsing this ToV. But it does seem plausible.
While the USSR was indeed able to exfiltrate secrets from Los Alamos to speed up its nuclear program, it took a few more years for it to actually develop a nuclear weapon.
Russell (and we don’t necessarily agree here) argued that the US could have established a monopoly on nuclear development through nuclear coercion. That strategy doesn’t have anything to do with preventing espionage.
Once the genie is out of the bottle, it doesn’t matter, does it? Much of China’s current tech achievements began with industrial espionage. You can’t constrain a game-changing technology while excluding espionage as a factor.
It’s exactly the same issue with AI.
While you have an interesting theoretical concept, there’s no way to derive a strategy from it that would lead to AI safety that I can see.
The idea for this particular theory of victory is that, if some country (for example, the US) develops TAI first, it could use TAI to prevent other countries (for example, China) from developing TAI as well — including via espionage.
If TAI grants a decisive strategic advantage, then it follows that such a monopoly could be effectively enforced (for example, it’s plausible that TAI-enabled cybersecurity would effectively protect against non-TAI cyberoffense).
Again, I’m not necessarily endorsing this ToV. But it does seem plausible.