Humans live with uncertainty and are accustomed to living with and acting in relativity.Computers work on absolutes.Humans make relations to objects and ideas by utilizing their senses and imagination.Computers have no reference.
Objects in language exist in context and reference to something else.Thus creating boundaries or groupings which computers can identify and a relationship matrix of some kind can be built upon that.
Thinking on the line of children.Why do they continue to ask why?Curiosity can only be built upon the boundaries of an established knowledge base.You can't ask what you don't know.
There are many ways to ask a question.Why didn't they win or why did they lose?How does a computer know which one to ask and what to base it on.There is also a problem with socializing.How would it learn to be social,build rapport,deal with sarcasm(Reminded of Sheldon here
),jokes and quotes.Also how would they deal with adjectives because they are very subjective.
Given a framework,an AI system can gain knowledge independently,comprehend it,solve problems but the true test lies in synthesizing knowledge independently.Lateral thinking would be interesting,like a philosophical AI.