Faces may not be the best example, because there's a lot of evidence babies are 'hardcoded' to be interested in and pay attention to faces. This implies there's

Faces may not be the best example, because there's a lot of evidence babies are 'hardcoded' to be interested in and pay attention to faces. This implies there's some knowledge of them built in. However, if you substitute abstract shapes (blocks, balls, etc.), your point absolutely stands up. In some ways, though, isn't it more interesting to have an intelligence that learns in a completely different way from humans, as opposed to one that learns through essentially the same method?