One example, continuing my previous post on FF, about why real AI β speaking in a general sense β cannot predict snow in summer.
In other words, it cannot make gross, fundamental mistakes if the methodology behind it is correct.
Imagine an AI used by a bank to evaluate the quality of a potential borrower.
As we know, banks were among the first to adopt AI-based systems for preliminary borrower assessment.
Could such an AI β one that has repeatedly and consistently predicted correctly, filtering out bad borrowers β make a mistake?
Of course it could. Rarely, perhaps extremely rarely β but could it happen?
But how?
Well, first of all, there is always the possibility of a statistical outlier. Something outside what has previously been observed, or something capable of producing emergent behavior.
But the much more important part is this "something" itself.
For example, with borrowers, we might encounter a fraudster. Someone who replaced documents, manipulated information, or used other obviously questionable but successful methods specifically designed to deceive the AI system.
And now back to the market.
This is roughly what we see happening here )
As well as a combination of rare factors β from the category of: who is weaker today, GBP or AUD?
And here comes the most important point.
A borrower may simply not repay the money and produce a loss.
The market, however, consists of a huge number of participants, which statistically makes such an absolute outcome much less likely.
Put simply β nobody gives up without a fight here )))
And this is exactly what we have been observing with great pleasure all this time )
I'm confident that this will be another such case )
