> "Proper machine learning systems would probably distinguish between someone doing something with a red flag only occasionally and a fanatic who does something based on a strong belief"
The point of the article was that even "proper" machine learning systems will generate either false positives or false negatives, either of which are damaging enough that they compromise the effectiveness of the approach.
The point of the article was that even "proper" machine learning systems will generate either false positives or false negatives, either of which are damaging enough that they compromise the effectiveness of the approach.