Improving AI with baby brains
Artificial intelligence researchers have constantly been seeking to make computers more like us with the ability to learn, hypothesise and make decisions. In the past, AI researchers have involved adults, but are now looking elsewhere for inspiration - the brain of a human child.
Researchers at the University of California, Berkely have been studying how children learn in the hopes of giving computers similar abilities by using exploratory and probabilistic reasoning tests. For example, given a choice between two biased jars with either pink or black lollipops, a lollipop was removed from each jar and hidden, so an infant could not see which color it was and thus have to rely on information from the jars about which Lollipop was the one they wanted. In most cases, the child crawlled to the lollipop from the mostly pink jar.