What happens if a flash crash analogous to that of the financial world, caused by AI algos, happens in strategic defense? (As AI is given more and more power in warfare...)

How is this avoidable?

(besides not using AI at all, which ofc won't happen)

Founder of @mercury - bringing cheap and scalable compute to AI developers 

Ironically, I doubt we can predict those risks (and how to mitigate) without leveraging AI-driven analytics. Too many plausible scenarios in which cause and correlation would be impossible for a human observer to determine… at least in real-time.