When training a neural network it’s extremely important to have data that’s pristine. If training data is labelled wrong it can permanently reinforce connections that will forever lead to incorrect decisions.
With this in mind I wonder why we still find it cute to lie to children. We teach them about Santa Clause, the Easter bunny, and the tooth fairy. They incorporate this data from their most trusted sources into their network which detects what’s true and what’s not. They grow up to be gullible, easily manipulated adults who believe in things like luck, horoscopes, ghosts, magic, and gods.
I put it to parents to recognize the damage they are doing to their children’s neural network by mislabeling training data. AI teaches us how to be better parents and you are not cruel for raising intelligent children.
from Artificial Intelligence http://ift.tt/2uDmVsn