> research has shown that children as young as 18 months can infer the goal of an adult's actions and develop alternate ways of reaching that goal themselves.
> In one example, infants saw an adult try to pull apart a barbell-shaped toy, but the adult failed to achieve that goal because the toy was stuck together and his hands slipped off the ends. The infants watched carefully and then decided to use alternate methods -- they wrapped their tiny fingers all the way around the ends and yanked especially hard -- duplicating what the adult intended to do.
I think the idea behind mimicry is goal identification (i.e.: see human wash a single dish, and identify the goal of 'getting all dishes clean'), not 'learn the way humans would perform this task and then do it'.
This passage from the article might help answer that:
"DeepMind learned to play video games by randomly taking any action it could. This may be fine for video games, but in the real world it could be expensive, time-consuming, and even deadly to have a robot trying to learn by trying every possible action to see what generated a “reward” and what didn’t. So Osaro also wrote algorithms that helps computers learn by mimicking humans."
