I don't think AI is a decisive strategic advantage, I think it is a decisive economic advantage.
A common overfit I see AGI thinkers make is to assume the DSA; if you actually look into the various possible military applications of frontier AI in eg 5 years time, none of them shake out as creating such a strong effect for the US vs a peer competitor country, so I think Dario should speak to a wider range of DC natsec folks. For the US vs anyone else, that was already lopsided anyway...
Examples: - EW - cyberoffense - underwater drones - social media disruption - NC3 integration - space-based monitoring & targeting - small-size drone swarms
Many of these shift the balance a little, but are equally constrained by other factors & bottlenecks, or retaliatory options. How strong each of these applications is is an exercise left to the reader (or Fable).
