Scientists are modelling artificial intelligence after baby brains. Why would they want to make computers similar to beings whose favourite pastimes are drooling and pooping? It makes perfect sense when you think about how malleable a baby’s grey matter is.
Artificially intelligent machines have a tough time with nuances and uncertainty. But babies, toddlers and preschoolers are great at interpreting such things. So Alison Gopnik, a developmental psychologist at UC Berkeley, and her colleague Tom Griffiths, are putting babies to the test to find ways to incorporate their abilities in to computer programming. “Children are the greatest learning machines in the universe,” Gopnik says. “Imagine if computers could learn as much and as quickly as they do.”
They’ve already found that at very young ages, babies can test hypotheses, detect statistical patterns and draw conclusions about important matters such as lollipops and toys — all the while adapting to changes.
As smart as computers are, youngsters can solve problems that machines can’t, including learning languages and interpreting causal relationships. If computers could be more like children, it might lead to digital tutoring programs, phone operators, or even robots that can identify genes associated with disease susceptibilities. The researchers are creating a centre at the Berkeley’s Institute of Human Development to meld baby and computer research.
And if an angry machine comes storming out of there one day in a baby robot rage, the good news is all you’ll need to do is find its binky. [ScienceBlog]