Machine learning algorithms are very commonly used for purposes mentioned in the article. But, one needs to understand that some widely used classifiers are bas

Machine learning algorithms are very commonly used for purposes mentioned in the article. But, one needs to understand that some widely used classifiers are based on probabilities of occurrence of all the features. I don't think such algorithms can ever completely replace human decision making especially for the purposes such as recruitment or university admissions, and it should not in my opinion. But, there are various other uses where there is quite a definitive code for decision making, machine learning algorithms can be used, in such situations it is more a problem of maximizing accuracy by different models of feature selection instead. The author also mentions a few glitches in the system, I'd say as there are more glitches, then we can use these glitches for training our models and so predicting such glitches can also be possible.