I understand the contention, but I don't think it's all that poor of a wording. In the end, who has more influence on the behavior of a neural network: The guy who wrote maybe a few dozen lines setting up its architecture, or the people who labeled the training data?

You should at least try to come up with an alternative headline that conveys to the average reader what is happening here, because "labeling data for machine learning" is not something that works either.