This is great idea!
It makes me think about the Chinese room argument from a new perspective: even if you are a human it is still hard to distinguish yourself from AI when the output channel is limited.
