# Isn't attention mask for BERT model useless?

I have just dived into deep learning for NLP, and now I'm learning how the BERT model works. What I found odd is why the BERT model needs to have an attention mask. As clearly shown in this tutorial https://huggingface.co/transformers/glossary.html:

from transformers import BertTokenizer
tokenizer = BertTokenizer.from_pretrained("bert-base-cased")

sequence_a = "This is a short sequence."
sequence_b = "This is a rather long sequence. It is at least longer than the sequence A."

encoded_sequence_a = tokenizer(sequence_a)["input_ids"]
encoded_sequence_b = tokenizer(sequence_b)["input_ids"]



Output of padded sequences input ids:

padded_sequences["input_ids"]

[[101, 1188, 1110, 170, 1603, 4954, 119, 102, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [101, 1188, 1110, 170, 1897, 1263, 4954, 119, 1135, 1110, 1120, 1655, 2039, 1190, 1103, 4954, 138, 119, 102]]


padded_sequences["attention_mask"]
[[1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]]


In the tutorial, it clearly states that an attention mask is needed to tell the model (BERT) which input ids need to be attended and which not (if an element in attention mask is 1 then the model will pay attention to that index, if it is 0 then model will not pay attention).

The thing I don't get is: why does BERT have an attention mask in the first place? Doesn't model need only input ids because you can clearly see that attention_mask has zeros on the same indices as the input_ids. Why does the model need to have an additional layer of difficulty added?

I know that BERT was created in google's "super duper laboratories", so I think the creators had something in their minds and had a strong reason for creating an attention mask as a part of the input.