Why AI cannot infer intent

AI cannot understand intent. (See Roger Schank's article - https://www.linkedin.com/pulse/can-computer-infer-intent-yet-elon-roger-schank/). AI not understanding intent is a special case of commonsense. Here is how - 

The relation between commonsense and English sentences is such that the meaning of the sentence emerges via commonsense interpreting something that is "incomplete" in the sentence. Consider this sentence - I love chocolates. Now, it is nowhere mentioned that he loves chocolates to eat. It could be to juggle them with his hands. But commonsense gives the correct interpretation that the meaning is "to eat". Similarly, if someone says that 'the US army defeated its enemy', most likely the enemy is the army of another country. (Actually the enemy could be anyone, say, one individual). So, here we see that commonsense completes the incomplete.

But intent is a special case of commonsense wherein there is nothing incomplete even in an explicit description. Yet the meaning and interpretation is left to commonsense judgment. Consider this sentence - John gave a ball to Jack. Here, if John gives a ball to Jack and one is asked to describe this action explicitly, all one would say - John gave a ball to Jack, or at the most, John gave a ball to Jack with his hands. But where is it captured in the English of the sentence that John had an intention to give the ball? Commonsense tells us that - John is a human being. He gave something. That is, he did some action. So there was an intent involved. But this commonsense is not the "completion of anything incomplete". Even the most explicit description of the action/event is not going to say something like - John gave a ball to Jack with intent. OR John had an intent of giving a ball to Jack and gave a ball to Jack.

This puts intent in a special class wherein explicit description too requires regular/typical commonsense judgment.