The more ambiguous problem you try to solve, the clearer you understand how your humand mind works; if you have enough consciousness.
I got this idea when I was looking at a picture taken in dark environment, I tried to tell what were the objects in the picture. Without carefully considering the picture closely, I would not tell anything from that. When I caught a line in the picture, I tried to figure out where the line was from, and I guess is it a human portrait or a car. A lot of process occurred there; guessing, matching, line correlation, bla bla. All happen within a blink, luckily, I was conscious enough to see my mind worked. On the other hand, I would not encounter this situation on any clear and explicit picture at all. That is because we take it for granted already on what we saw and the brain process is so fast that we could not catch what part is “input” or “interpretation” or “feeling” or “response”. Thanks to the ambiguousness, I understand my mind better. Hopefully, I think I could exploit this idea on machine learning especially in object recognition.