AI Training Data: Why 99% Accuracy Isn't Enough! #shorts

To illustrate this, if you have 1 million events per day and you know one in 100,000 you see a malicious event and you have a detective a verifier which has 99% of true positive rate and only 1% of false positive rate which is exceptional right what do you think how many how good is this um this oracle and verifier and it comes down 10,000 false positives so if you train now your AI model based on this detector It's not going to be good. It's not going to generize on this bad stuff. The precision is going to be very very bad.