You can make an educated guess if you would understand the intricacies of the programming. In this case, it's most likely blurting out words and phrases that statistically most adequately fit the (perhaps somewhat leading) questions.