I came across a very interesting question:

Many people now believe that AI can analyze competitions, analyze trades, analyze market trends, but when it comes to truly big upsets and major market movements, it can't predict them; humans are still needed.
A game suddenly produces an extreme score, silver jumps from double digits to 80, crude oil suddenly surges.
If AI only predicts based on historical data and past distributions, it is very likely to keep reminding you: deviation is too large, risk is too high, take profits, revert back.
But real big market moves often happen when historical experience starts to fail. So the core issue is that current AI, when facing out-of-distribution extreme events, lacks sufficient generalization, adaptability, and framework reconstruction capabilities.
It can calculate probabilities, perform risk control, and organize information, but it’s very hard for it to judge: is this just an abnormal fluctuation, or the start of a new cycle, a new narrative, or a new pricing system?
However, most humans also struggle to hold onto big market trends, and are even more easily influenced by fear, greed, and emotions.
Humans’ true advantage may lie in rebuilding understanding frameworks during extreme changes.