,,It's easy to make a statement correct by making it vague. '' What's funny is that the last dataset that Google built for evaluating dialogues has exactly the

,,It's easy to make a statement correct by making it vague. '' What's funny is that the last dataset that Google built for evaluating dialogues has exactly these 2 metrics: correctness and specificity.