Built-in regularization, probably, plus cross-validation. These techniques aren't new; they're included in a number of ML libraries already - just not at this level of automation.

Probably the same way a human would do it: By splitting the dataset into training, testing and validation buckets.

Probably some rule based expert system to analyze results . Looking for overfitting could be done by creating validity set. So you would have test, train, and validity .