you don't have to be ML only - it can be Human lifeguard augmented by ML Yep, when it comes to such things my background leans towards NLP. Not sure what the O

you don't have to be ML only - it can be Human lifeguard augmented by ML Yep, when it comes to such things my background leans towards NLP. Not sure what the OCR state of the art is these days, but it used to be side-by-side human-assisted computer: The computer did the bulk of the work. For language translation (at least in lower stakes applications) it was the reverse: A human assisted by a computer. Even within computer-only NLP methods, they often benefited from a mixed-model approach using both rule-based & stochastic methods together.