Abstract
Today, many people have been paying academic and journalistic attention to the so-called machine learning or deep learning. But in many cases, there is a kind of serious lack of awareness of the ontological meanings behind these methods. In deep learning in which a kind of approximate calculation using discrete values is used, a set of concepts such as learning, trial-and-error, minimalizing of errors would disappear. In this article, we will critically and ethically examine how various meanings would disappear or appear in mathematics, machine learning or in our activity with our own imagination and views on life.