Advances in AI areas like computer vision seem to mostly come from academics who publish their work. Once an innovation is published, it is available for all to use or imitate. I don't see how a state actor is going to make some major AI advance that it is also able to conceal from other states.<p>I'd be particularly bearish on AI for grand strategy. If LLMs need billions of examples of text to outperform humans, an analogous war strategy AI would need to be trained on more examples of warfare than have been conducted in human history.