Why use RNN for name entity recognition?

I don’t get why I should use an RNN if I would like to find the name “Harry” in the sentence provided by Andrew.

Concretely, Andrew gave this example: “Harry Potter and Hermione Granger invented a new spell”. Then he went on explaining that we can use an RNN in order to find out if the first word in the sentence is, in fact, Harry.

Why use RNN rather than a comparison like “Harry” = “Harry” => TRUE? I am lost. Apparently, I misunderstood something.
Please explain :slight_smile:

It’s more than just finding a specific word. It’s also about finding associations between sequences of words.